Datasets:

snappr

/

idefics-food-ai

like 0

Follow

Snappr 18

[ "\n### CONTEXT\n\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n4. A set of one or more reference images of the same menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, I need you to review the image and assign a category to it.\n\nThe review result can only be one of three options: `fail`, `soft_fail`, or `pass`.\n\nThe review assigned to the input image should be a result of an evaluation process that involves different criteria like:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Accuracy: How well the image adjusts to the given description and the reference images.\n - Realism and coherence of the photo (the background makes sense, no absurd elements, no elements with weird shapes, etc.)\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be similar to the reference images. We don't want too much food.\n - Correct ingredients: The image should show the correct ingredients for the menu item.\n\n### MENU ITEM NAME:\nSPICY EDAMAME.\n\n### MENU ITEM DESCRIPTION:\nSautéed with spicy tsuyu no moto sauce..\n\nConsidering everything above, analyze and review the input image. I need your answer to be just the category that you think corresponds to the\nimage you will reviewing (fail, soft_fail, or pass). Don't return anything else.\n\nI repeat, ONLY return the category.\n ", "\n### CONTEXT\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n4. A set of one or more reference images of the same menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, and the images below, I need you to review the image and try to find any possible issues with it.\n\nThe analysis of the input image should be a result of an evaluation process that involves different criteria like:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Accuracy: How well the image adjusts to the given description and the reference images.\n - Realism and coherence of the photo (the background makes sense, no absurd elements, no elements with weird shapes, etc.)\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be similar to the reference images. We don't want too much food.\n - Correct ingredients: The image should show the correct ingredients for the menu item.\n\n### MENU ITEM NAME:\nSPICY EDAMAME.\n\n### MENU ITEM DESCRIPTION:\nSautéed with spicy tsuyu no moto sauce..\n\nConsidering everything above, analyze and review the input image.\nI am going to give you a list of four issues that you connsider are preventing the image to pass a QA process.\nSelect the issue that best applies to the input image.\n\nThis is the list of possible issues: [other, food-arrangement-too-similar-to-reference-image, ingredient-shape-incorrect, ingredient-texture-incorrect]\n\nReturn only the text from the issue that best applies to the input image.\n ", "\nI am going to give you an image of a menu item, which may be a food dish, a side or a beverage.\n\nThis input image was generated using AI. I am going to give you a list of four possible description for this menu item, separated by \" | \".\nSelect the one that best describes the image.\n\nThis is the list of possible descriptions: [Sautéed with spicy tsuyu no moto sauce. | Thinly sliced lamb and beef cooked on a vertical spit, served over a bed of seasoned fries. | Large side of fries served with ketchup and ranch.]\n\nReturn only the text from the list that best describes the menu item.\n ", "\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, and the images below, I need you to review the image and try to find any possible aesthetics-related issues with it.\n\nPlease consider the following criteria:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be reasonable. We don't want too much or too little food.\n\n### MENU ITEM NAME:\nSPICY EDAMAME.\n\n### MENU ITEM DESCRIPTION:\nSautéed with spicy tsuyu no moto sauce..\n\nConsidering everything above, analyze and review the input image.\nRespond only stating if the input image has any possible aesthetic issues or not.\n\nOnly possible responses are \"The image is good. It contains no significant aesthetic-related issues\" or \"The image has significant aesthetic issues that should be addressed\".\n" ]

[ "soft_fail", "From the list of possible issues, the one that best applies to the input image is other.", "The text that best describes the menu item is \"Sautéed with spicy tsuyu no moto sauce.\".", "The image has significant aesthetic issues that should be addressed." ]

[]

[ "other" ]

[]

[ "\n### CONTEXT\n\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n4. A set of one or more reference images of the same menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, I need you to review the image and assign a category to it.\n\nThe review result can only be one of three options: `fail`, `soft_fail`, or `pass`.\n\nThe review assigned to the input image should be a result of an evaluation process that involves different criteria like:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Accuracy: How well the image adjusts to the given description and the reference images.\n - Realism and coherence of the photo (the background makes sense, no absurd elements, no elements with weird shapes, etc.)\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be similar to the reference images. We don't want too much food.\n - Correct ingredients: The image should show the correct ingredients for the menu item.\n\n### MENU ITEM NAME:\nDeep Fried Mushrooms.\n\n### MENU ITEM DESCRIPTION:\n<No description provided.>.\n\nConsidering everything above, analyze and review the input image. I need your answer to be just the category that you think corresponds to the\nimage you will reviewing (fail, soft_fail, or pass). Don't return anything else.\n\nI repeat, ONLY return the category.\n ", "\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n4. A set of one or more reference images of the same menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, I need you to review the image and tray to find any possible issues with it.\n\nThe analysis of the input image should be a result of an evaluation process that involves different criteria like:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Accuracy: How well the image adjusts to the given description and the reference images.\n - Realism and coherence of the photo (the background makes sense, no absurd elements, no elements with weird shapes, etc.)\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be similar to the reference images. We don't want too much food.\n - Correct ingredients: The image should show the correct ingredients for the menu item.\n\n### MENU ITEM NAME:\nDeep Fried Mushrooms.\n\n### MENU ITEM DESCRIPTION:\n<No description provided.>.\n\nConsidering everything above, analyze and review the input image.\nI need your answer to be a list of one or more issues that you connsider are preventing the image to pass a QA process.\nIf you consider that the image has no significant issues and is ok to be approved, return \"No issues\".\n ", "\n### CONTEXT\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n4. A set of one or more reference images of the same menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, and the images below, I need you to review the image and try to find any possible issues with it.\n\nThe analysis of the input image should be a result of an evaluation process that involves different criteria like:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Accuracy: How well the image adjusts to the given description and the reference images.\n - Realism and coherence of the photo (the background makes sense, no absurd elements, no elements with weird shapes, etc.)\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be similar to the reference images. We don't want too much food.\n - Correct ingredients: The image should show the correct ingredients for the menu item.\n\n### MENU ITEM NAME:\nDeep Fried Mushrooms.\n\n### MENU ITEM DESCRIPTION:\n<No description provided.>.\n\nConsidering everything above, analyze and review the input image.\nI am going to give you a list of four issues that you connsider are preventing the image to pass a QA process.\nSelect the issue that best applies to the input image.\n\nThis is the list of possible issues: [20-padding-from-edge-of-food, too-blurry, ingredients-size, portion-size-incorrect]\n\nReturn only the text from the issue that best applies to the input image.\n ", "\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, and the images below, I need you to review the image and try to find any possible aesthetics-related issues with it.\n\nPlease consider the following criteria:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be reasonable. We don't want too much or too little food.\n\n### MENU ITEM NAME:\nDeep Fried Mushrooms.\n\n### MENU ITEM DESCRIPTION:\n<No description provided.>.\n\nConsidering everything above, analyze and review the input image.\nRespond only stating if the input image has any possible aesthetic issues or not.\n\nOnly possible responses are \"The image is good. It contains no significant aesthetic-related issues\" or \"The image has significant aesthetic issues that should be addressed\".\n" ]

[ "pass", "The image is of a good quality and should pass the QA process. No issues", "None of the possible issues applies to the input image.", "The image is good. It contains no significant aesthetic-related issues." ]

[]

[]

[]

[ "\n### CONTEXT\n\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n4. A set of one or more reference images of the same menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, I need you to review the image and assign a category to it.\n\nThe review result can only be one of three options: `fail`, `soft_fail`, or `pass`.\n\nThe review assigned to the input image should be a result of an evaluation process that involves different criteria like:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Accuracy: How well the image adjusts to the given description and the reference images.\n - Realism and coherence of the photo (the background makes sense, no absurd elements, no elements with weird shapes, etc.)\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be similar to the reference images. We don't want too much food.\n - Correct ingredients: The image should show the correct ingredients for the menu item.\n\n### MENU ITEM NAME:\nCheeseburger.\n\n### MENU ITEM DESCRIPTION:\nChoice of sharp cheddar, Monterey Jack, American or Swiss cheese. All burgers are half a pound and include lettuce, tomato, onion mayo, and pickle. All the burgers will be cooked medium unless specified. Choice of one of the following French fries, mashed potatoes and gravy or organic spring mix salad..\n\nConsidering everything above, analyze and review the input image. I need your answer to be just the category that you think corresponds to the\nimage you will reviewing (fail, soft_fail, or pass). Don't return anything else.\n\nI repeat, ONLY return the category.\n ", "\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n4. A set of one or more reference images of the same menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, I need you to review the image and tray to find any possible issues with it.\n\nThe analysis of the input image should be a result of an evaluation process that involves different criteria like:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Accuracy: How well the image adjusts to the given description and the reference images.\n - Realism and coherence of the photo (the background makes sense, no absurd elements, no elements with weird shapes, etc.)\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be similar to the reference images. We don't want too much food.\n - Correct ingredients: The image should show the correct ingredients for the menu item.\n\n### MENU ITEM NAME:\nCheeseburger.\n\n### MENU ITEM DESCRIPTION:\nChoice of sharp cheddar, Monterey Jack, American or Swiss cheese. All burgers are half a pound and include lettuce, tomato, onion mayo, and pickle. All the burgers will be cooked medium unless specified. Choice of one of the following French fries, mashed potatoes and gravy or organic spring mix salad..\n\nConsidering everything above, analyze and review the input image.\nI need your answer to be a list of one or more issues that you connsider are preventing the image to pass a QA process.\nIf you consider that the image has no significant issues and is ok to be approved, return \"No issues\".\n ", "\n### CONTEXT\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n4. A set of one or more reference images of the same menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, and the images below, I need you to review the image and try to find any possible issues with it.\n\nThe analysis of the input image should be a result of an evaluation process that involves different criteria like:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Accuracy: How well the image adjusts to the given description and the reference images.\n - Realism and coherence of the photo (the background makes sense, no absurd elements, no elements with weird shapes, etc.)\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be similar to the reference images. We don't want too much food.\n - Correct ingredients: The image should show the correct ingredients for the menu item.\n\n### MENU ITEM NAME:\nCheeseburger.\n\n### MENU ITEM DESCRIPTION:\nChoice of sharp cheddar, Monterey Jack, American or Swiss cheese. All burgers are half a pound and include lettuce, tomato, onion mayo, and pickle. All the burgers will be cooked medium unless specified. Choice of one of the following French fries, mashed potatoes and gravy or organic spring mix salad..\n\nConsidering everything above, analyze and review the input image.\nI am going to give you a list of four issues that you connsider are preventing the image to pass a QA process.\nSelect the issue that best applies to the input image.\n\nThis is the list of possible issues: [ingredients-type, portion-amount, container, ingredient-shape-incorrect]\n\nReturn only the text from the issue that best applies to the input image.\n ", "\nI am going to give you an image of a menu item, which may be a food dish, a side or a beverage.\n\nThis input image was generated using AI. I am going to give you a list of four possible description for this menu item, separated by \" | \".\nSelect the one that best describes the image.\n\nThis is the list of possible descriptions: [Choice of sharp cheddar, Monterey Jack, American or Swiss cheese. All burgers are half a pound and include lettuce, tomato, onion mayo, and pickle. All the burgers will be cooked medium unless specified. Choice of one of the following French fries, mashed potatoes and gravy or organic spring mix salad. | 28g protein. Banana, almond butter, almond milk, Earthfusion protein, granola. | ½ LB fresh meat, our house made sauce, Avocado, Your Choice of Cheese, Lettuce, Tomatoes, Pickles and onions]\n\nReturn only the text from the list that best describes the menu item.\n ", "\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, and the images below, I need you to review the image and try to find any possible aesthetics-related issues with it.\n\nPlease consider the following criteria:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be reasonable. We don't want too much or too little food.\n\n### MENU ITEM NAME:\nCheeseburger.\n\n### MENU ITEM DESCRIPTION:\nChoice of sharp cheddar, Monterey Jack, American or Swiss cheese. All burgers are half a pound and include lettuce, tomato, onion mayo, and pickle. All the burgers will be cooked medium unless specified. Choice of one of the following French fries, mashed potatoes and gravy or organic spring mix salad..\n\nConsidering everything above, analyze and review the input image.\nRespond only stating if the input image has any possible aesthetic issues or not.\n\nOnly possible responses are \"The image is good. It contains no significant aesthetic-related issues\" or \"The image has significant aesthetic issues that should be addressed\".\n" ]

[ "fail", "container", "From the list of possible issues, the one that best applies to the input image is container.", "The text that best describes the menu item is \"Choice of sharp cheddar, Monterey Jack, American or Swiss cheese. All burgers are half a pound and include lettuce, tomato, onion mayo, and pickle. All the burgers will be cooked medium unless specified. Choice of one of the following French fries, mashed potatoes and gravy or organic spring mix salad.\".", "The image has significant aesthetic issues that should be addressed." ]

[]

[ "container" ]

[]

[ "\n### CONTEXT\n\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n4. A set of one or more reference images of the same menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, I need you to review the image and assign a category to it.\n\nThe review result can only be one of three options: `fail`, `soft_fail`, or `pass`.\n\nThe review assigned to the input image should be a result of an evaluation process that involves different criteria like:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Accuracy: How well the image adjusts to the given description and the reference images.\n - Realism and coherence of the photo (the background makes sense, no absurd elements, no elements with weird shapes, etc.)\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be similar to the reference images. We don't want too much food.\n - Correct ingredients: The image should show the correct ingredients for the menu item.\n\n### MENU ITEM NAME:\nJarritos Mandarin.\n\n### MENU ITEM DESCRIPTION:\n [Cal 140].\n\nConsidering everything above, analyze and review the input image. I need your answer to be just the category that you think corresponds to the\nimage you will reviewing (fail, soft_fail, or pass). Don't return anything else.\n\nI repeat, ONLY return the category.\n ", "\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n4. A set of one or more reference images of the same menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, I need you to review the image and tray to find any possible issues with it.\n\nThe analysis of the input image should be a result of an evaluation process that involves different criteria like:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Accuracy: How well the image adjusts to the given description and the reference images.\n - Realism and coherence of the photo (the background makes sense, no absurd elements, no elements with weird shapes, etc.)\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be similar to the reference images. We don't want too much food.\n - Correct ingredients: The image should show the correct ingredients for the menu item.\n\n### MENU ITEM NAME:\nJarritos Mandarin.\n\n### MENU ITEM DESCRIPTION:\n [Cal 140].\n\nConsidering everything above, analyze and review the input image.\nI need your answer to be a list of one or more issues that you connsider are preventing the image to pass a QA process.\nIf you consider that the image has no significant issues and is ok to be approved, return \"No issues\".\n ", "\n### CONTEXT\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n4. A set of one or more reference images of the same menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, and the images below, I need you to review the image and try to find any possible issues with it.\n\nThe analysis of the input image should be a result of an evaluation process that involves different criteria like:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Accuracy: How well the image adjusts to the given description and the reference images.\n - Realism and coherence of the photo (the background makes sense, no absurd elements, no elements with weird shapes, etc.)\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be similar to the reference images. We don't want too much food.\n - Correct ingredients: The image should show the correct ingredients for the menu item.\n\n### MENU ITEM NAME:\nJarritos Mandarin.\n\n### MENU ITEM DESCRIPTION:\n [Cal 140].\n\nConsidering everything above, analyze and review the input image.\nI am going to give you a list of four issues that you connsider are preventing the image to pass a QA process.\nSelect the issue that best applies to the input image.\n\nThis is the list of possible issues: [scale-incorrect, realistic-plating-arrangement, scale, shadows-too-tight]\n\nReturn only the text from the issue that best applies to the input image.\n ", "\nI am going to give you an image of a menu item, which may be a food dish, a side or a beverage.\n\nThis input image was generated using AI. I am going to give you a list of four possible description for this menu item, separated by \" | \".\nSelect the one that best describes the image.\n\nThis is the list of possible descriptions: [ [Cal 140] | Sandwich served with combo of falafel & beef gyro. Served with one white sauce and one red sauce. | Gluten-Free. Basmati rice cooked with lamb.]\n\nReturn only the text from the list that best describes the menu item.\n ", "\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, and the images below, I need you to review the image and try to find any possible aesthetics-related issues with it.\n\nPlease consider the following criteria:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be reasonable. We don't want too much or too little food.\n\n### MENU ITEM NAME:\nJarritos Mandarin.\n\n### MENU ITEM DESCRIPTION:\n [Cal 140].\n\nConsidering everything above, analyze and review the input image.\nRespond only stating if the input image has any possible aesthetic issues or not.\n\nOnly possible responses are \"The image is good. It contains no significant aesthetic-related issues\" or \"The image has significant aesthetic issues that should be addressed\".\n" ]

[ "soft_fail", "realistic-plating-arrangement", "From the list of possible issues, the one that best applies to the input image is realistic-plating-arrangement.", "The text that best describes the menu item is \" [Cal 140]\".", "The image has significant aesthetic issues that should be addressed." ]

[]

[ "realistic-plating-arrangement" ]

[]

[ "\n### CONTEXT\n\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n4. A set of one or more reference images of the same menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, I need you to review the image and assign a category to it.\n\nThe review result can only be one of three options: `fail`, `soft_fail`, or `pass`.\n\nThe review assigned to the input image should be a result of an evaluation process that involves different criteria like:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Accuracy: How well the image adjusts to the given description and the reference images.\n - Realism and coherence of the photo (the background makes sense, no absurd elements, no elements with weird shapes, etc.)\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be similar to the reference images. We don't want too much food.\n - Correct ingredients: The image should show the correct ingredients for the menu item.\n\n### MENU ITEM NAME:\n5a. Chicken Tikka Masala.\n\n### MENU ITEM DESCRIPTION:\nBoneless Chicken cooked in tomato sauce, yogurt, onions and spices..\n\nConsidering everything above, analyze and review the input image. I need your answer to be just the category that you think corresponds to the\nimage you will reviewing (fail, soft_fail, or pass). Don't return anything else.\n\nI repeat, ONLY return the category.\n ", "\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n4. A set of one or more reference images of the same menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, I need you to review the image and tray to find any possible issues with it.\n\nThe analysis of the input image should be a result of an evaluation process that involves different criteria like:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Accuracy: How well the image adjusts to the given description and the reference images.\n - Realism and coherence of the photo (the background makes sense, no absurd elements, no elements with weird shapes, etc.)\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be similar to the reference images. We don't want too much food.\n - Correct ingredients: The image should show the correct ingredients for the menu item.\n\n### MENU ITEM NAME:\n5a. Chicken Tikka Masala.\n\n### MENU ITEM DESCRIPTION:\nBoneless Chicken cooked in tomato sauce, yogurt, onions and spices..\n\nConsidering everything above, analyze and review the input image.\nI need your answer to be a list of one or more issues that you connsider are preventing the image to pass a QA process.\nIf you consider that the image has no significant issues and is ok to be approved, return \"No issues\".\n ", "\n### CONTEXT\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n4. A set of one or more reference images of the same menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, and the images below, I need you to review the image and try to find any possible issues with it.\n\nThe analysis of the input image should be a result of an evaluation process that involves different criteria like:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Accuracy: How well the image adjusts to the given description and the reference images.\n - Realism and coherence of the photo (the background makes sense, no absurd elements, no elements with weird shapes, etc.)\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be similar to the reference images. We don't want too much food.\n - Correct ingredients: The image should show the correct ingredients for the menu item.\n\n### MENU ITEM NAME:\n5a. Chicken Tikka Masala.\n\n### MENU ITEM DESCRIPTION:\nBoneless Chicken cooked in tomato sauce, yogurt, onions and spices..\n\nConsidering everything above, analyze and review the input image.\nI am going to give you a list of four issues that you connsider are preventing the image to pass a QA process.\nSelect the issue that best applies to the input image.\n\nThis is the list of possible issues: [ingredient-color-incorrect, blurry, ingredient-texture-too-mushy, shadows-missing]\n\nReturn only the text from the issue that best applies to the input image.\n ", "\nI am going to give you an image of a menu item, which may be a food dish, a side or a beverage.\n\nThis input image was generated using AI. I am going to give you a list of four possible description for this menu item, separated by \" | \".\nSelect the one that best describes the image.\n\nThis is the list of possible descriptions: [Freshly cooked basmati rice flavored with cumin and turmeric | Strawberry, blueberry, banana, blue agave, and Oat milk. | Boneless Chicken cooked in tomato sauce, yogurt, onions and spices.]\n\nReturn only the text from the list that best describes the menu item.\n ", "\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, and the images below, I need you to review the image and try to find any possible aesthetics-related issues with it.\n\nPlease consider the following criteria:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be reasonable. We don't want too much or too little food.\n\n### MENU ITEM NAME:\n5a. Chicken Tikka Masala.\n\n### MENU ITEM DESCRIPTION:\nBoneless Chicken cooked in tomato sauce, yogurt, onions and spices..\n\nConsidering everything above, analyze and review the input image.\nRespond only stating if the input image has any possible aesthetic issues or not.\n\nOnly possible responses are \"The image is good. It contains no significant aesthetic-related issues\" or \"The image has significant aesthetic issues that should be addressed\".\n" ]

[ "pass", "ingredient-color-incorrect", "From the list of possible issues, the one that best applies to the input image is ingredient-color-incorrect.", "The text that best describes the menu item is \"Boneless Chicken cooked in tomato sauce, yogurt, onions and spices.\".", "The image is good. It contains no significant aesthetic-related issues." ]

[ "ingredient-color-incorrect" ]

[]

[]

[ "\n### CONTEXT\n\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n4. A set of one or more reference images of the same menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, I need you to review the image and assign a category to it.\n\nThe review result can only be one of three options: `fail`, `soft_fail`, or `pass`.\n\nThe review assigned to the input image should be a result of an evaluation process that involves different criteria like:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Accuracy: How well the image adjusts to the given description and the reference images.\n - Realism and coherence of the photo (the background makes sense, no absurd elements, no elements with weird shapes, etc.)\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be similar to the reference images. We don't want too much food.\n - Correct ingredients: The image should show the correct ingredients for the menu item.\n\n### MENU ITEM NAME:\nFish Masala.\n\n### MENU ITEM DESCRIPTION:\nTilapia fish marinated in lime juice, ginger, garlic, and Indian spices. It is then cooked in onion tomato masala based gravy. Finally it’s tempered with black peppers, garlic, curry leaves, mustard seeds, and garnished with fresh cilantro..\n\nConsidering everything above, analyze and review the input image. I need your answer to be just the category that you think corresponds to the\nimage you will reviewing (fail, soft_fail, or pass). Don't return anything else.\n\nI repeat, ONLY return the category.\n ", "\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n4. A set of one or more reference images of the same menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, I need you to review the image and tray to find any possible issues with it.\n\nThe analysis of the input image should be a result of an evaluation process that involves different criteria like:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Accuracy: How well the image adjusts to the given description and the reference images.\n - Realism and coherence of the photo (the background makes sense, no absurd elements, no elements with weird shapes, etc.)\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be similar to the reference images. We don't want too much food.\n - Correct ingredients: The image should show the correct ingredients for the menu item.\n\n### MENU ITEM NAME:\nFish Masala.\n\n### MENU ITEM DESCRIPTION:\nTilapia fish marinated in lime juice, ginger, garlic, and Indian spices. It is then cooked in onion tomato masala based gravy. Finally it’s tempered with black peppers, garlic, curry leaves, mustard seeds, and garnished with fresh cilantro..\n\nConsidering everything above, analyze and review the input image.\nI need your answer to be a list of one or more issues that you connsider are preventing the image to pass a QA process.\nIf you consider that the image has no significant issues and is ok to be approved, return \"No issues\".\n ", "\n### CONTEXT\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n4. A set of one or more reference images of the same menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, and the images below, I need you to review the image and try to find any possible issues with it.\n\nThe analysis of the input image should be a result of an evaluation process that involves different criteria like:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Accuracy: How well the image adjusts to the given description and the reference images.\n - Realism and coherence of the photo (the background makes sense, no absurd elements, no elements with weird shapes, etc.)\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be similar to the reference images. We don't want too much food.\n - Correct ingredients: The image should show the correct ingredients for the menu item.\n\n### MENU ITEM NAME:\nFish Masala.\n\n### MENU ITEM DESCRIPTION:\nTilapia fish marinated in lime juice, ginger, garlic, and Indian spices. It is then cooked in onion tomato masala based gravy. Finally it’s tempered with black peppers, garlic, curry leaves, mustard seeds, and garnished with fresh cilantro..\n\nConsidering everything above, analyze and review the input image.\nI am going to give you a list of four issues that you connsider are preventing the image to pass a QA process.\nSelect the issue that best applies to the input image.\n\nThis is the list of possible issues: [general-oversaturation, ingredients---unnatural-blending-, indistinct-from-other-final-image(s), ingredient-color-incorrect]\n\nReturn only the text from the issue that best applies to the input image.\n ", "\nI am going to give you an image of a menu item, which may be a food dish, a side or a beverage.\n\nThis input image was generated using AI. I am going to give you a list of four possible description for this menu item, separated by \" | \".\nSelect the one that best describes the image.\n\nThis is the list of possible descriptions: [Calzone stuffed with chicken, onions, and a drizzle of BBQ sauce, all enveloped in cheese and marinara goodness. | Bite-sized Mini Churros dusted with cinnamon sugar. Perfect for sharing, with yourself. | Tilapia fish marinated in lime juice, ginger, garlic, and Indian spices. It is then cooked in onion tomato masala based gravy. Finally it’s tempered with black peppers, garlic, curry leaves, mustard seeds, and garnished with fresh cilantro.]\n\nReturn only the text from the list that best describes the menu item.\n ", "\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, and the images below, I need you to review the image and try to find any possible aesthetics-related issues with it.\n\nPlease consider the following criteria:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be reasonable. We don't want too much or too little food.\n\n### MENU ITEM NAME:\nFish Masala.\n\n### MENU ITEM DESCRIPTION:\nTilapia fish marinated in lime juice, ginger, garlic, and Indian spices. It is then cooked in onion tomato masala based gravy. Finally it’s tempered with black peppers, garlic, curry leaves, mustard seeds, and garnished with fresh cilantro..\n\nConsidering everything above, analyze and review the input image.\nRespond only stating if the input image has any possible aesthetic issues or not.\n\nOnly possible responses are \"The image is good. It contains no significant aesthetic-related issues\" or \"The image has significant aesthetic issues that should be addressed\".\n" ]

[ "soft_fail", "ingredient-color-incorrect, unrealistic-placement-of-ingredients", "From the list of possible issues, the one that best applies to the input image is ingredient-color-incorrect.", "The text that best describes the menu item is \"Tilapia fish marinated in lime juice, ginger, garlic, and Indian spices. It is then cooked in onion tomato masala based gravy. Finally it’s tempered with black peppers, garlic, curry leaves, mustard seeds, and garnished with fresh cilantro.\".", "The image has significant aesthetic issues that should be addressed." ]

[ "ingredient-color-incorrect" ]

[ "unrealistic-placement-of-ingredients" ]

[]

[ "\n### CONTEXT\n\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n4. A set of one or more reference images of the same menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, I need you to review the image and assign a category to it.\n\nThe review result can only be one of three options: `fail`, `soft_fail`, or `pass`.\n\nThe review assigned to the input image should be a result of an evaluation process that involves different criteria like:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Accuracy: How well the image adjusts to the given description and the reference images.\n - Realism and coherence of the photo (the background makes sense, no absurd elements, no elements with weird shapes, etc.)\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be similar to the reference images. We don't want too much food.\n - Correct ingredients: The image should show the correct ingredients for the menu item.\n\n### MENU ITEM NAME:\n#16 Mike's Chicken Philly.\n\n### MENU ITEM DESCRIPTION:\nGrilled onions, peppers & white American cheese..\n\nConsidering everything above, analyze and review the input image. I need your answer to be just the category that you think corresponds to the\nimage you will reviewing (fail, soft_fail, or pass). Don't return anything else.\n\nI repeat, ONLY return the category.\n ", "\n### CONTEXT\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n4. A set of one or more reference images of the same menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, and the images below, I need you to review the image and try to find any possible issues with it.\n\nThe analysis of the input image should be a result of an evaluation process that involves different criteria like:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Accuracy: How well the image adjusts to the given description and the reference images.\n - Realism and coherence of the photo (the background makes sense, no absurd elements, no elements with weird shapes, etc.)\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be similar to the reference images. We don't want too much food.\n - Correct ingredients: The image should show the correct ingredients for the menu item.\n\n### MENU ITEM NAME:\n#16 Mike's Chicken Philly.\n\n### MENU ITEM DESCRIPTION:\nGrilled onions, peppers & white American cheese..\n\nConsidering everything above, analyze and review the input image.\nI am going to give you a list of four issues that you connsider are preventing the image to pass a QA process.\nSelect the issue that best applies to the input image.\n\nThis is the list of possible issues: [white-balance-incorrect, highlights-too-dark, other, general-underexposure]\n\nReturn only the text from the issue that best applies to the input image.\n ", "\nI am going to give you an image of a menu item, which may be a food dish, a side or a beverage.\n\nThis input image was generated using AI. I am going to give you a list of four possible description for this menu item, separated by \" | \".\nSelect the one that best describes the image.\n\nThis is the list of possible descriptions: [Your choice of protein stir fried with dried bean curd, bean sprout, egg, scallion and sweet turnip, topped with fresh toasted peanut. | Carbonated mineral water, served in a bottle. | Grilled onions, peppers & white American cheese.]\n\nReturn only the text from the list that best describes the menu item.\n ", "\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, and the images below, I need you to review the image and try to find any possible aesthetics-related issues with it.\n\nPlease consider the following criteria:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be reasonable. We don't want too much or too little food.\n\n### MENU ITEM NAME:\n#16 Mike's Chicken Philly.\n\n### MENU ITEM DESCRIPTION:\nGrilled onions, peppers & white American cheese..\n\nConsidering everything above, analyze and review the input image.\nRespond only stating if the input image has any possible aesthetic issues or not.\n\nOnly possible responses are \"The image is good. It contains no significant aesthetic-related issues\" or \"The image has significant aesthetic issues that should be addressed\".\n" ]

[ "fail", "From the list of possible issues, the one that best applies to the input image is other.", "The text that best describes the menu item is \"Grilled onions, peppers & white American cheese.\".", "The image is good. It contains no significant aesthetic-related issues." ]

[]

[]

[ "other" ]

[ "\n### CONTEXT\n\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n4. A set of one or more reference images of the same menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, I need you to review the image and assign a category to it.\n\nThe review result can only be one of three options: `fail`, `soft_fail`, or `pass`.\n\nThe review assigned to the input image should be a result of an evaluation process that involves different criteria like:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Accuracy: How well the image adjusts to the given description and the reference images.\n - Realism and coherence of the photo (the background makes sense, no absurd elements, no elements with weird shapes, etc.)\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be similar to the reference images. We don't want too much food.\n - Correct ingredients: The image should show the correct ingredients for the menu item.\n\n### MENU ITEM NAME:\nKAIYO Roll.\n\n### MENU ITEM DESCRIPTION:\nCanadian red crab, salmon, tempura asparagus, avocado, rocoto, aioli, onion, jalapeño, garlic chili oil, crispy garlic.\n\nConsidering everything above, analyze and review the input image. I need your answer to be just the category that you think corresponds to the\nimage you will reviewing (fail, soft_fail, or pass). Don't return anything else.\n\nI repeat, ONLY return the category.\n ", "\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n4. A set of one or more reference images of the same menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, I need you to review the image and tray to find any possible issues with it.\n\nThe analysis of the input image should be a result of an evaluation process that involves different criteria like:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Accuracy: How well the image adjusts to the given description and the reference images.\n - Realism and coherence of the photo (the background makes sense, no absurd elements, no elements with weird shapes, etc.)\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be similar to the reference images. We don't want too much food.\n - Correct ingredients: The image should show the correct ingredients for the menu item.\n\n### MENU ITEM NAME:\nKAIYO Roll.\n\n### MENU ITEM DESCRIPTION:\nCanadian red crab, salmon, tempura asparagus, avocado, rocoto, aioli, onion, jalapeño, garlic chili oil, crispy garlic.\n\nConsidering everything above, analyze and review the input image.\nI need your answer to be a list of one or more issues that you connsider are preventing the image to pass a QA process.\nIf you consider that the image has no significant issues and is ok to be approved, return \"No issues\".\n ", "\n### CONTEXT\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n4. A set of one or more reference images of the same menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, and the images below, I need you to review the image and try to find any possible issues with it.\n\nThe analysis of the input image should be a result of an evaluation process that involves different criteria like:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Accuracy: How well the image adjusts to the given description and the reference images.\n - Realism and coherence of the photo (the background makes sense, no absurd elements, no elements with weird shapes, etc.)\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be similar to the reference images. We don't want too much food.\n - Correct ingredients: The image should show the correct ingredients for the menu item.\n\n### MENU ITEM NAME:\nKAIYO Roll.\n\n### MENU ITEM DESCRIPTION:\nCanadian red crab, salmon, tempura asparagus, avocado, rocoto, aioli, onion, jalapeño, garlic chili oil, crispy garlic.\n\nConsidering everything above, analyze and review the input image.\nI am going to give you a list of four issues that you connsider are preventing the image to pass a QA process.\nSelect the issue that best applies to the input image.\n\nThis is the list of possible issues: [uneven-spread-of-light, container-incorrect, editing-artifacts-visible, ingredients-size]\n\nReturn only the text from the issue that best applies to the input image.\n ", "\nI am going to give you an image of a menu item, which may be a food dish, a side or a beverage.\n\nThis input image was generated using AI. I am going to give you a list of four possible description for this menu item, separated by \" | \".\nSelect the one that best describes the image.\n\nThis is the list of possible descriptions: [Canadian red crab, salmon, tempura asparagus, avocado, rocoto, aioli, onion, jalapeño, garlic chili oil, crispy garlic | Spicy ground chicken, toasted sesame,cilantro, green onion, organic dashi egg, spice bomb | 22 oz ice-cold Coca-Cola® Zero Sugar]\n\nReturn only the text from the list that best describes the menu item.\n ", "\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, and the images below, I need you to review the image and try to find any possible aesthetics-related issues with it.\n\nPlease consider the following criteria:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be reasonable. We don't want too much or too little food.\n\n### MENU ITEM NAME:\nKAIYO Roll.\n\n### MENU ITEM DESCRIPTION:\nCanadian red crab, salmon, tempura asparagus, avocado, rocoto, aioli, onion, jalapeño, garlic chili oil, crispy garlic.\n\nConsidering everything above, analyze and review the input image.\nRespond only stating if the input image has any possible aesthetic issues or not.\n\nOnly possible responses are \"The image is good. It contains no significant aesthetic-related issues\" or \"The image has significant aesthetic issues that should be addressed\".\n" ]

[ "fail", "unit-quantity-incorrect, ingredient-shape-incorrect, ingredient-texture-incorrect, ingredient-color-incorrect, container-incorrect", "From the list of possible issues, the one that best applies to the input image is container-incorrect.", "The text that best describes the menu item is \"Canadian red crab, salmon, tempura asparagus, avocado, rocoto, aioli, onion, jalapeño, garlic chili oil, crispy garlic\".", "The image has significant aesthetic issues that should be addressed." ]

[ "unit-quantity-incorrect", "ingredient-shape-incorrect", "ingredient-texture-incorrect", "ingredient-color-incorrect" ]

[ "container-incorrect" ]

[]

[ "\n### CONTEXT\n\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n4. A set of one or more reference images of the same menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, I need you to review the image and assign a category to it.\n\nThe review result can only be one of three options: `fail`, `soft_fail`, or `pass`.\n\nThe review assigned to the input image should be a result of an evaluation process that involves different criteria like:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Accuracy: How well the image adjusts to the given description and the reference images.\n - Realism and coherence of the photo (the background makes sense, no absurd elements, no elements with weird shapes, etc.)\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be similar to the reference images. We don't want too much food.\n - Correct ingredients: The image should show the correct ingredients for the menu item.\n\n### MENU ITEM NAME:\nChicken Bacon Ranch.\n\n### MENU ITEM DESCRIPTION:\nRanch, chicken, bacon, mozzarella, and provolone cheese..\n\nConsidering everything above, analyze and review the input image. I need your answer to be just the category that you think corresponds to the\nimage you will reviewing (fail, soft_fail, or pass). Don't return anything else.\n\nI repeat, ONLY return the category.\n ", "\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n4. A set of one or more reference images of the same menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, I need you to review the image and tray to find any possible issues with it.\n\nThe analysis of the input image should be a result of an evaluation process that involves different criteria like:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Accuracy: How well the image adjusts to the given description and the reference images.\n - Realism and coherence of the photo (the background makes sense, no absurd elements, no elements with weird shapes, etc.)\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be similar to the reference images. We don't want too much food.\n - Correct ingredients: The image should show the correct ingredients for the menu item.\n\n### MENU ITEM NAME:\nChicken Bacon Ranch.\n\n### MENU ITEM DESCRIPTION:\nRanch, chicken, bacon, mozzarella, and provolone cheese..\n\nConsidering everything above, analyze and review the input image.\nI need your answer to be a list of one or more issues that you connsider are preventing the image to pass a QA process.\nIf you consider that the image has no significant issues and is ok to be approved, return \"No issues\".\n ", "\n### CONTEXT\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n4. A set of one or more reference images of the same menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, and the images below, I need you to review the image and try to find any possible issues with it.\n\nThe analysis of the input image should be a result of an evaluation process that involves different criteria like:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Accuracy: How well the image adjusts to the given description and the reference images.\n - Realism and coherence of the photo (the background makes sense, no absurd elements, no elements with weird shapes, etc.)\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be similar to the reference images. We don't want too much food.\n - Correct ingredients: The image should show the correct ingredients for the menu item.\n\n### MENU ITEM NAME:\nChicken Bacon Ranch.\n\n### MENU ITEM DESCRIPTION:\nRanch, chicken, bacon, mozzarella, and provolone cheese..\n\nConsidering everything above, analyze and review the input image.\nI am going to give you a list of four issues that you connsider are preventing the image to pass a QA process.\nSelect the issue that best applies to the input image.\n\nThis is the list of possible issues: [ingredients-texture, quantity-of-units-too-many, incorrect-perspective, unit-quantity-incorrect]\n\nReturn only the text from the issue that best applies to the input image.\n ", "\nI am going to give you an image of a menu item, which may be a food dish, a side or a beverage.\n\nThis input image was generated using AI. I am going to give you a list of four possible description for this menu item, separated by \" | \".\nSelect the one that best describes the image.\n\nThis is the list of possible descriptions: [Pepperoni Pizza Slice | Ranch, chicken, bacon, mozzarella, and provolone cheese. | Crispy chicken, buttermilk waffle, bacon butter syrup]\n\nReturn only the text from the list that best describes the menu item.\n ", "\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, and the images below, I need you to review the image and try to find any possible aesthetics-related issues with it.\n\nPlease consider the following criteria:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be reasonable. We don't want too much or too little food.\n\n### MENU ITEM NAME:\nChicken Bacon Ranch.\n\n### MENU ITEM DESCRIPTION:\nRanch, chicken, bacon, mozzarella, and provolone cheese..\n\nConsidering everything above, analyze and review the input image.\nRespond only stating if the input image has any possible aesthetic issues or not.\n\nOnly possible responses are \"The image is good. It contains no significant aesthetic-related issues\" or \"The image has significant aesthetic issues that should be addressed\".\n" ]

[ "fail", "ingredients-texture", "From the list of possible issues, the one that best applies to the input image is ingredients-texture.", "The text that best describes the menu item is \"Ranch, chicken, bacon, mozzarella, and provolone cheese.\".", "The image is good. It contains no significant aesthetic-related issues." ]

[ "ingredients-texture" ]

[]

[]

[ "\n### CONTEXT\n\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n4. A set of one or more reference images of the same menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, I need you to review the image and assign a category to it.\n\nThe review result can only be one of three options: `fail`, `soft_fail`, or `pass`.\n\nThe review assigned to the input image should be a result of an evaluation process that involves different criteria like:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Accuracy: How well the image adjusts to the given description and the reference images.\n - Realism and coherence of the photo (the background makes sense, no absurd elements, no elements with weird shapes, etc.)\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be similar to the reference images. We don't want too much food.\n - Correct ingredients: The image should show the correct ingredients for the menu item.\n\n### MENU ITEM NAME:\n2-Liter Soda.\n\n### MENU ITEM DESCRIPTION:\nServed cold..\n\nConsidering everything above, analyze and review the input image. I need your answer to be just the category that you think corresponds to the\nimage you will reviewing (fail, soft_fail, or pass). Don't return anything else.\n\nI repeat, ONLY return the category.\n ", "\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n4. A set of one or more reference images of the same menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, I need you to review the image and tray to find any possible issues with it.\n\nThe analysis of the input image should be a result of an evaluation process that involves different criteria like:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Accuracy: How well the image adjusts to the given description and the reference images.\n - Realism and coherence of the photo (the background makes sense, no absurd elements, no elements with weird shapes, etc.)\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be similar to the reference images. We don't want too much food.\n - Correct ingredients: The image should show the correct ingredients for the menu item.\n\n### MENU ITEM NAME:\n2-Liter Soda.\n\n### MENU ITEM DESCRIPTION:\nServed cold..\n\nConsidering everything above, analyze and review the input image.\nI need your answer to be a list of one or more issues that you connsider are preventing the image to pass a QA process.\nIf you consider that the image has no significant issues and is ok to be approved, return \"No issues\".\n ", "\n### CONTEXT\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n4. A set of one or more reference images of the same menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, and the images below, I need you to review the image and try to find any possible issues with it.\n\nThe analysis of the input image should be a result of an evaluation process that involves different criteria like:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Accuracy: How well the image adjusts to the given description and the reference images.\n - Realism and coherence of the photo (the background makes sense, no absurd elements, no elements with weird shapes, etc.)\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be similar to the reference images. We don't want too much food.\n - Correct ingredients: The image should show the correct ingredients for the menu item.\n\n### MENU ITEM NAME:\n2-Liter Soda.\n\n### MENU ITEM DESCRIPTION:\nServed cold..\n\nConsidering everything above, analyze and review the input image.\nI am going to give you a list of four issues that you connsider are preventing the image to pass a QA process.\nSelect the issue that best applies to the input image.\n\nThis is the list of possible issues: [container-incorrect, lightning-and-shadow, other, oversaturated]\n\nReturn only the text from the issue that best applies to the input image.\n ", "\nI am going to give you an image of a menu item, which may be a food dish, a side or a beverage.\n\nThis input image was generated using AI. I am going to give you a list of four possible description for this menu item, separated by \" | \".\nSelect the one that best describes the image.\n\nThis is the list of possible descriptions: [Served cold. | Eight slices. | Boneless cubed chicken skewered and cooked in a clay oven.]\n\nReturn only the text from the list that best describes the menu item.\n ", "\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, and the images below, I need you to review the image and try to find any possible aesthetics-related issues with it.\n\nPlease consider the following criteria:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be reasonable. We don't want too much or too little food.\n\n### MENU ITEM NAME:\n2-Liter Soda.\n\n### MENU ITEM DESCRIPTION:\nServed cold..\n\nConsidering everything above, analyze and review the input image.\nRespond only stating if the input image has any possible aesthetic issues or not.\n\nOnly possible responses are \"The image is good. It contains no significant aesthetic-related issues\" or \"The image has significant aesthetic issues that should be addressed\".\n" ]

[ "fail", "ingredients-type, other", "From the list of possible issues, the one that best applies to the input image is other.", "The text that best describes the menu item is \"Served cold.\".", "The image is good. It contains no significant aesthetic-related issues." ]

[ "ingredients-type", "other" ]

[]

[]

[ "\n### CONTEXT\n\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n4. A set of one or more reference images of the same menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, I need you to review the image and assign a category to it.\n\nThe review result can only be one of three options: `fail`, `soft_fail`, or `pass`.\n\nThe review assigned to the input image should be a result of an evaluation process that involves different criteria like:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Accuracy: How well the image adjusts to the given description and the reference images.\n - Realism and coherence of the photo (the background makes sense, no absurd elements, no elements with weird shapes, etc.)\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be similar to the reference images. We don't want too much food.\n - Correct ingredients: The image should show the correct ingredients for the menu item.\n\n### MENU ITEM NAME:\nLinguine con Vongole.\n\n### MENU ITEM DESCRIPTION:\nManila clams, zucchini, garlic white wine sauce..\n\nConsidering everything above, analyze and review the input image. I need your answer to be just the category that you think corresponds to the\nimage you will reviewing (fail, soft_fail, or pass). Don't return anything else.\n\nI repeat, ONLY return the category.\n ", "\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n4. A set of one or more reference images of the same menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, I need you to review the image and tray to find any possible issues with it.\n\nThe analysis of the input image should be a result of an evaluation process that involves different criteria like:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Accuracy: How well the image adjusts to the given description and the reference images.\n - Realism and coherence of the photo (the background makes sense, no absurd elements, no elements with weird shapes, etc.)\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be similar to the reference images. We don't want too much food.\n - Correct ingredients: The image should show the correct ingredients for the menu item.\n\n### MENU ITEM NAME:\nLinguine con Vongole.\n\n### MENU ITEM DESCRIPTION:\nManila clams, zucchini, garlic white wine sauce..\n\nConsidering everything above, analyze and review the input image.\nI need your answer to be a list of one or more issues that you connsider are preventing the image to pass a QA process.\nIf you consider that the image has no significant issues and is ok to be approved, return \"No issues\".\n ", "\n### CONTEXT\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n4. A set of one or more reference images of the same menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, and the images below, I need you to review the image and try to find any possible issues with it.\n\nThe analysis of the input image should be a result of an evaluation process that involves different criteria like:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Accuracy: How well the image adjusts to the given description and the reference images.\n - Realism and coherence of the photo (the background makes sense, no absurd elements, no elements with weird shapes, etc.)\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be similar to the reference images. We don't want too much food.\n - Correct ingredients: The image should show the correct ingredients for the menu item.\n\n### MENU ITEM NAME:\nLinguine con Vongole.\n\n### MENU ITEM DESCRIPTION:\nManila clams, zucchini, garlic white wine sauce..\n\nConsidering everything above, analyze and review the input image.\nI am going to give you a list of four issues that you connsider are preventing the image to pass a QA process.\nSelect the issue that best applies to the input image.\n\nThis is the list of possible issues: [portion-amount, scale-incorrect, editing-artifacts-visible, white-balance-incorrect]\n\nReturn only the text from the issue that best applies to the input image.\n ", "\nI am going to give you an image of a menu item, which may be a food dish, a side or a beverage.\n\nThis input image was generated using AI. I am going to give you a list of four possible description for this menu item, separated by \" | \".\nSelect the one that best describes the image.\n\nThis is the list of possible descriptions: [Manila clams, zucchini, garlic white wine sauce. | Sharp cheddar cheese, bacon, and bbq sauce. All burgers are half a pound and include lettuce, tomato, onion mayo, and pickle. All the burgers will be cooked medium unless specified. Choice of one of the following French fries, mashed potatoes and gravy or organic spring mix salad. | Our Signature Dish!! Gobi Manchurian is a popular Indo-Chinese dish where crispy cauliflower florets are tossed in a manchurian sauce.]\n\nReturn only the text from the list that best describes the menu item.\n ", "\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, and the images below, I need you to review the image and try to find any possible aesthetics-related issues with it.\n\nPlease consider the following criteria:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be reasonable. We don't want too much or too little food.\n\n### MENU ITEM NAME:\nLinguine con Vongole.\n\n### MENU ITEM DESCRIPTION:\nManila clams, zucchini, garlic white wine sauce..\n\nConsidering everything above, analyze and review the input image.\nRespond only stating if the input image has any possible aesthetic issues or not.\n\nOnly possible responses are \"The image is good. It contains no significant aesthetic-related issues\" or \"The image has significant aesthetic issues that should be addressed\".\n" ]

[ "fail", "ingredients-type, portion-amount, ingredients-size, ingredients-texture", "From the list of possible issues, the one that best applies to the input image is portion-amount.", "The text that best describes the menu item is \"Manila clams, zucchini, garlic white wine sauce.\".", "The image is good. It contains no significant aesthetic-related issues." ]

[ "ingredients-type", "portion-amount", "ingredients-size", "ingredients-texture" ]

[]

[]

[ "\n### CONTEXT\n\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n4. A set of one or more reference images of the same menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, I need you to review the image and assign a category to it.\n\nThe review result can only be one of three options: `fail`, `soft_fail`, or `pass`.\n\nThe review assigned to the input image should be a result of an evaluation process that involves different criteria like:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Accuracy: How well the image adjusts to the given description and the reference images.\n - Realism and coherence of the photo (the background makes sense, no absurd elements, no elements with weird shapes, etc.)\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be similar to the reference images. We don't want too much food.\n - Correct ingredients: The image should show the correct ingredients for the menu item.\n\n### MENU ITEM NAME:\nVegetable Biryani.\n\n### MENU ITEM DESCRIPTION:\nSpiced seasoned vegetables cooked with Indian spices and basmati rice. Served with house raita..\n\nConsidering everything above, analyze and review the input image. I need your answer to be just the category that you think corresponds to the\nimage you will reviewing (fail, soft_fail, or pass). Don't return anything else.\n\nI repeat, ONLY return the category.\n ", "\n### CONTEXT\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n4. A set of one or more reference images of the same menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, and the images below, I need you to review the image and try to find any possible issues with it.\n\nThe analysis of the input image should be a result of an evaluation process that involves different criteria like:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Accuracy: How well the image adjusts to the given description and the reference images.\n - Realism and coherence of the photo (the background makes sense, no absurd elements, no elements with weird shapes, etc.)\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be similar to the reference images. We don't want too much food.\n - Correct ingredients: The image should show the correct ingredients for the menu item.\n\n### MENU ITEM NAME:\nVegetable Biryani.\n\n### MENU ITEM DESCRIPTION:\nSpiced seasoned vegetables cooked with Indian spices and basmati rice. Served with house raita..\n\nConsidering everything above, analyze and review the input image.\nI am going to give you a list of four issues that you connsider are preventing the image to pass a QA process.\nSelect the issue that best applies to the input image.\n\nThis is the list of possible issues: [portion-of-image-is-too-similar-to-other-parts, other, container-too-small, ingredient-size-incorrect]\n\nReturn only the text from the issue that best applies to the input image.\n ", "\nI am going to give you an image of a menu item, which may be a food dish, a side or a beverage.\n\nThis input image was generated using AI. I am going to give you a list of four possible description for this menu item, separated by \" | \".\nSelect the one that best describes the image.\n\nThis is the list of possible descriptions: [Spiced seasoned vegetables cooked with Indian spices and basmati rice. Served with house raita. | Vanilla and chocolate layers alternating with fresh whipped cream and topped with a rich fudge icing. | Fried octopus ball.]\n\nReturn only the text from the list that best describes the menu item.\n ", "\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, and the images below, I need you to review the image and try to find any possible aesthetics-related issues with it.\n\nPlease consider the following criteria:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be reasonable. We don't want too much or too little food.\n\n### MENU ITEM NAME:\nVegetable Biryani.\n\n### MENU ITEM DESCRIPTION:\nSpiced seasoned vegetables cooked with Indian spices and basmati rice. Served with house raita..\n\nConsidering everything above, analyze and review the input image.\nRespond only stating if the input image has any possible aesthetic issues or not.\n\nOnly possible responses are \"The image is good. It contains no significant aesthetic-related issues\" or \"The image has significant aesthetic issues that should be addressed\".\n" ]

[ "fail", "From the list of possible issues, the one that best applies to the input image is other.", "The text that best describes the menu item is \"Spiced seasoned vegetables cooked with Indian spices and basmati rice. Served with house raita.\".", "The image is good. It contains no significant aesthetic-related issues." ]

[ "other" ]

[]

[]

[ "\n### CONTEXT\n\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n4. A set of one or more reference images of the same menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, I need you to review the image and assign a category to it.\n\nThe review result can only be one of three options: `fail`, `soft_fail`, or `pass`.\n\nThe review assigned to the input image should be a result of an evaluation process that involves different criteria like:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Accuracy: How well the image adjusts to the given description and the reference images.\n - Realism and coherence of the photo (the background makes sense, no absurd elements, no elements with weird shapes, etc.)\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be similar to the reference images. We don't want too much food.\n - Correct ingredients: The image should show the correct ingredients for the menu item.\n\n### MENU ITEM NAME:\nTrio Spreads.\n\n### MENU ITEM DESCRIPTION:\nMuhammara, Baba Ganoush, Tzatziki & warm pita.\n\nConsidering everything above, analyze and review the input image. I need your answer to be just the category that you think corresponds to the\nimage you will reviewing (fail, soft_fail, or pass). Don't return anything else.\n\nI repeat, ONLY return the category.\n ", "\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n4. A set of one or more reference images of the same menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, I need you to review the image and tray to find any possible issues with it.\n\nThe analysis of the input image should be a result of an evaluation process that involves different criteria like:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Accuracy: How well the image adjusts to the given description and the reference images.\n - Realism and coherence of the photo (the background makes sense, no absurd elements, no elements with weird shapes, etc.)\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be similar to the reference images. We don't want too much food.\n - Correct ingredients: The image should show the correct ingredients for the menu item.\n\n### MENU ITEM NAME:\nTrio Spreads.\n\n### MENU ITEM DESCRIPTION:\nMuhammara, Baba Ganoush, Tzatziki & warm pita.\n\nConsidering everything above, analyze and review the input image.\nI need your answer to be a list of one or more issues that you connsider are preventing the image to pass a QA process.\nIf you consider that the image has no significant issues and is ok to be approved, return \"No issues\".\n ", "\n### CONTEXT\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n4. A set of one or more reference images of the same menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, and the images below, I need you to review the image and try to find any possible issues with it.\n\nThe analysis of the input image should be a result of an evaluation process that involves different criteria like:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Accuracy: How well the image adjusts to the given description and the reference images.\n - Realism and coherence of the photo (the background makes sense, no absurd elements, no elements with weird shapes, etc.)\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be similar to the reference images. We don't want too much food.\n - Correct ingredients: The image should show the correct ingredients for the menu item.\n\n### MENU ITEM NAME:\nTrio Spreads.\n\n### MENU ITEM DESCRIPTION:\nMuhammara, Baba Ganoush, Tzatziki & warm pita.\n\nConsidering everything above, analyze and review the input image.\nI am going to give you a list of four issues that you connsider are preventing the image to pass a QA process.\nSelect the issue that best applies to the input image.\n\nThis is the list of possible issues: [shadows-too-small, general-wb-temperature-too-warm, other, scale]\n\nReturn only the text from the issue that best applies to the input image.\n ", "\nI am going to give you an image of a menu item, which may be a food dish, a side or a beverage.\n\nThis input image was generated using AI. I am going to give you a list of four possible description for this menu item, separated by \" | \".\nSelect the one that best describes the image.\n\nThis is the list of possible descriptions: [Muhammara, Baba Ganoush, Tzatziki & warm pita | Sliced potatoes topped with Ocopa sauce (yellow ají pepper, peanuts & huacatay cream sauce). Served with egg & Peruvian olives. | Cheese Pizza]\n\nReturn only the text from the list that best describes the menu item.\n ", "\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, and the images below, I need you to review the image and try to find any possible aesthetics-related issues with it.\n\nPlease consider the following criteria:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be reasonable. We don't want too much or too little food.\n\n### MENU ITEM NAME:\nTrio Spreads.\n\n### MENU ITEM DESCRIPTION:\nMuhammara, Baba Ganoush, Tzatziki & warm pita.\n\nConsidering everything above, analyze and review the input image.\nRespond only stating if the input image has any possible aesthetic issues or not.\n\nOnly possible responses are \"The image is good. It contains no significant aesthetic-related issues\" or \"The image has significant aesthetic issues that should be addressed\".\n" ]

[ "fail", "ingredients-size, ingredients-texture, container, scale", "From the list of possible issues, the one that best applies to the input image is scale.", "The text that best describes the menu item is \"Muhammara, Baba Ganoush, Tzatziki & warm pita\".", "The image has significant aesthetic issues that should be addressed." ]

[ "ingredients-size", "ingredients-texture" ]

[ "container", "scale" ]

[]

[ "\n### CONTEXT\n\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n4. A set of one or more reference images of the same menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, I need you to review the image and assign a category to it.\n\nThe review result can only be one of three options: `fail`, `soft_fail`, or `pass`.\n\nThe review assigned to the input image should be a result of an evaluation process that involves different criteria like:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Accuracy: How well the image adjusts to the given description and the reference images.\n - Realism and coherence of the photo (the background makes sense, no absurd elements, no elements with weird shapes, etc.)\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be similar to the reference images. We don't want too much food.\n - Correct ingredients: The image should show the correct ingredients for the menu item.\n\n### MENU ITEM NAME:\nSuper Soft Brioche Loaf.\n\n### MENU ITEM DESCRIPTION:\nOur delicious line of brioche breads are rich and golden with a thin crust and a Super Soft and silky inside. We use a special mixing process that requires the slow and gradual addition of softened butter, responsible for the taste and texture..\n\nConsidering everything above, analyze and review the input image. I need your answer to be just the category that you think corresponds to the\nimage you will reviewing (fail, soft_fail, or pass). Don't return anything else.\n\nI repeat, ONLY return the category.\n ", "\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n4. A set of one or more reference images of the same menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, I need you to review the image and tray to find any possible issues with it.\n\nThe analysis of the input image should be a result of an evaluation process that involves different criteria like:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Accuracy: How well the image adjusts to the given description and the reference images.\n - Realism and coherence of the photo (the background makes sense, no absurd elements, no elements with weird shapes, etc.)\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be similar to the reference images. We don't want too much food.\n - Correct ingredients: The image should show the correct ingredients for the menu item.\n\n### MENU ITEM NAME:\nSuper Soft Brioche Loaf.\n\n### MENU ITEM DESCRIPTION:\nOur delicious line of brioche breads are rich and golden with a thin crust and a Super Soft and silky inside. We use a special mixing process that requires the slow and gradual addition of softened butter, responsible for the taste and texture..\n\nConsidering everything above, analyze and review the input image.\nI need your answer to be a list of one or more issues that you connsider are preventing the image to pass a QA process.\nIf you consider that the image has no significant issues and is ok to be approved, return \"No issues\".\n ", "\n### CONTEXT\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n4. A set of one or more reference images of the same menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, and the images below, I need you to review the image and try to find any possible issues with it.\n\nThe analysis of the input image should be a result of an evaluation process that involves different criteria like:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Accuracy: How well the image adjusts to the given description and the reference images.\n - Realism and coherence of the photo (the background makes sense, no absurd elements, no elements with weird shapes, etc.)\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be similar to the reference images. We don't want too much food.\n - Correct ingredients: The image should show the correct ingredients for the menu item.\n\n### MENU ITEM NAME:\nSuper Soft Brioche Loaf.\n\n### MENU ITEM DESCRIPTION:\nOur delicious line of brioche breads are rich and golden with a thin crust and a Super Soft and silky inside. We use a special mixing process that requires the slow and gradual addition of softened butter, responsible for the taste and texture..\n\nConsidering everything above, analyze and review the input image.\nI am going to give you a list of four issues that you connsider are preventing the image to pass a QA process.\nSelect the issue that best applies to the input image.\n\nThis is the list of possible issues: [serving-size-too-little, container-too-big, ingredient-texture-incorrect, food-arrangement-too-similar-to-reference-image]\n\nReturn only the text from the issue that best applies to the input image.\n ", "\nI am going to give you an image of a menu item, which may be a food dish, a side or a beverage.\n\nThis input image was generated using AI. I am going to give you a list of four possible description for this menu item, separated by \" | \".\nSelect the one that best describes the image.\n\nThis is the list of possible descriptions: [Our delicious line of brioche breads are rich and golden with a thin crust and a Super Soft and silky inside. We use a special mixing process that requires the slow and gradual addition of softened butter, responsible for the taste and texture. | Mushrooms, gorgonzola, and walnuts in a light cream sauce. | Coconut jelly & pearl.]\n\nReturn only the text from the list that best describes the menu item.\n ", "\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, and the images below, I need you to review the image and try to find any possible aesthetics-related issues with it.\n\nPlease consider the following criteria:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be reasonable. We don't want too much or too little food.\n\n### MENU ITEM NAME:\nSuper Soft Brioche Loaf.\n\n### MENU ITEM DESCRIPTION:\nOur delicious line of brioche breads are rich and golden with a thin crust and a Super Soft and silky inside. We use a special mixing process that requires the slow and gradual addition of softened butter, responsible for the taste and texture..\n\nConsidering everything above, analyze and review the input image.\nRespond only stating if the input image has any possible aesthetic issues or not.\n\nOnly possible responses are \"The image is good. It contains no significant aesthetic-related issues\" or \"The image has significant aesthetic issues that should be addressed\".\n" ]

[ "fail", "shadows-missing, food-arrangement-too-similar-to-reference-image, pixel-by-pixel-copy-and-paste, portion-of-image-is-too-similar-to-other-parts", "From the list of possible issues, the one that best applies to the input image is food-arrangement-too-similar-to-reference-image.", "The text that best describes the menu item is \"Our delicious line of brioche breads are rich and golden with a thin crust and a Super Soft and silky inside. We use a special mixing process that requires the slow and gradual addition of softened butter, responsible for the taste and texture.\".", "The image has significant aesthetic issues that should be addressed." ]

[]

[ "shadows-missing" ]

[ "food-arrangement-too-similar-to-reference-image", "pixel-by-pixel-copy-and-paste", "portion-of-image-is-too-similar-to-other-parts" ]

[ "\n### CONTEXT\n\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n4. A set of one or more reference images of the same menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, I need you to review the image and assign a category to it.\n\nThe review result can only be one of three options: `fail`, `soft_fail`, or `pass`.\n\nThe review assigned to the input image should be a result of an evaluation process that involves different criteria like:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Accuracy: How well the image adjusts to the given description and the reference images.\n - Realism and coherence of the photo (the background makes sense, no absurd elements, no elements with weird shapes, etc.)\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be similar to the reference images. We don't want too much food.\n - Correct ingredients: The image should show the correct ingredients for the menu item.\n\n### MENU ITEM NAME:\nCinnamon & Honey Fried Chicken.\n\n### MENU ITEM DESCRIPTION:\nSpiced vanilla ice cream, bits of fried chicken croissant.\nContains: Milk, Wheat, Soy.\n\nConsidering everything above, analyze and review the input image. I need your answer to be just the category that you think corresponds to the\nimage you will reviewing (fail, soft_fail, or pass). Don't return anything else.\n\nI repeat, ONLY return the category.\n ", "\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n4. A set of one or more reference images of the same menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, I need you to review the image and tray to find any possible issues with it.\n\nThe analysis of the input image should be a result of an evaluation process that involves different criteria like:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Accuracy: How well the image adjusts to the given description and the reference images.\n - Realism and coherence of the photo (the background makes sense, no absurd elements, no elements with weird shapes, etc.)\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be similar to the reference images. We don't want too much food.\n - Correct ingredients: The image should show the correct ingredients for the menu item.\n\n### MENU ITEM NAME:\nCinnamon & Honey Fried Chicken.\n\n### MENU ITEM DESCRIPTION:\nSpiced vanilla ice cream, bits of fried chicken croissant.\nContains: Milk, Wheat, Soy.\n\nConsidering everything above, analyze and review the input image.\nI need your answer to be a list of one or more issues that you connsider are preventing the image to pass a QA process.\nIf you consider that the image has no significant issues and is ok to be approved, return \"No issues\".\n ", "\n### CONTEXT\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n4. A set of one or more reference images of the same menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, and the images below, I need you to review the image and try to find any possible issues with it.\n\nThe analysis of the input image should be a result of an evaluation process that involves different criteria like:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Accuracy: How well the image adjusts to the given description and the reference images.\n - Realism and coherence of the photo (the background makes sense, no absurd elements, no elements with weird shapes, etc.)\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be similar to the reference images. We don't want too much food.\n - Correct ingredients: The image should show the correct ingredients for the menu item.\n\n### MENU ITEM NAME:\nCinnamon & Honey Fried Chicken.\n\n### MENU ITEM DESCRIPTION:\nSpiced vanilla ice cream, bits of fried chicken croissant.\nContains: Milk, Wheat, Soy.\n\nConsidering everything above, analyze and review the input image.\nI am going to give you a list of four issues that you connsider are preventing the image to pass a QA process.\nSelect the issue that best applies to the input image.\n\nThis is the list of possible issues: [quantity-of-units-too-many, ingredients-type, container-too-small, container-too-big]\n\nReturn only the text from the issue that best applies to the input image.\n ", "\nI am going to give you an image of a menu item, which may be a food dish, a side or a beverage.\n\nThis input image was generated using AI. I am going to give you a list of four possible description for this menu item, separated by \" | \".\nSelect the one that best describes the image.\n\nThis is the list of possible descriptions: [House-made with ketchup, mayo, yellow mustard, sweet relish, Worcestershire sauce and Tabasco sauce – made to Gott’s secret proportions! 12 oz jar. | Spiced vanilla ice cream, bits of fried chicken croissant.\nContains: Milk, Wheat, Soy | Homemade low-carb brownie made with almond flour and organic ’Monkfruit’ sweetener. Gluten-free. *Contains nuts. Net carb: 1g.]\n\nReturn only the text from the list that best describes the menu item.\n ", "\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, and the images below, I need you to review the image and try to find any possible aesthetics-related issues with it.\n\nPlease consider the following criteria:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be reasonable. We don't want too much or too little food.\n\n### MENU ITEM NAME:\nCinnamon & Honey Fried Chicken.\n\n### MENU ITEM DESCRIPTION:\nSpiced vanilla ice cream, bits of fried chicken croissant.\nContains: Milk, Wheat, Soy.\n\nConsidering everything above, analyze and review the input image.\nRespond only stating if the input image has any possible aesthetic issues or not.\n\nOnly possible responses are \"The image is good. It contains no significant aesthetic-related issues\" or \"The image has significant aesthetic issues that should be addressed\".\n" ]

[ "fail", "ingredients-type, ingredients-color, ingredients-texture, ingredients-shape", "From the list of possible issues, the one that best applies to the input image is ingredients-type.", "The text that best describes the menu item is \"Spiced vanilla ice cream, bits of fried chicken croissant.\nContains: Milk, Wheat, Soy\".", "The image is good. It contains no significant aesthetic-related issues." ]

[ "ingredients-type", "ingredients-color", "ingredients-texture", "ingredients-shape" ]

[]

[]

[ "\n### CONTEXT\n\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n4. A set of one or more reference images of the same menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, I need you to review the image and assign a category to it.\n\nThe review result can only be one of three options: `fail`, `soft_fail`, or `pass`.\n\nThe review assigned to the input image should be a result of an evaluation process that involves different criteria like:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Accuracy: How well the image adjusts to the given description and the reference images.\n - Realism and coherence of the photo (the background makes sense, no absurd elements, no elements with weird shapes, etc.)\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be similar to the reference images. We don't want too much food.\n - Correct ingredients: The image should show the correct ingredients for the menu item.\n\n### MENU ITEM NAME:\nHouse Salad.\n\n### MENU ITEM DESCRIPTION:\nLettuce, tomatoes, onions, cilantro and house dressing..\n\nConsidering everything above, analyze and review the input image. I need your answer to be just the category that you think corresponds to the\nimage you will reviewing (fail, soft_fail, or pass). Don't return anything else.\n\nI repeat, ONLY return the category.\n ", "\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n4. A set of one or more reference images of the same menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, I need you to review the image and tray to find any possible issues with it.\n\nThe analysis of the input image should be a result of an evaluation process that involves different criteria like:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Accuracy: How well the image adjusts to the given description and the reference images.\n - Realism and coherence of the photo (the background makes sense, no absurd elements, no elements with weird shapes, etc.)\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be similar to the reference images. We don't want too much food.\n - Correct ingredients: The image should show the correct ingredients for the menu item.\n\n### MENU ITEM NAME:\nHouse Salad.\n\n### MENU ITEM DESCRIPTION:\nLettuce, tomatoes, onions, cilantro and house dressing..\n\nConsidering everything above, analyze and review the input image.\nI need your answer to be a list of one or more issues that you connsider are preventing the image to pass a QA process.\nIf you consider that the image has no significant issues and is ok to be approved, return \"No issues\".\n ", "\n### CONTEXT\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n4. A set of one or more reference images of the same menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, and the images below, I need you to review the image and try to find any possible issues with it.\n\nThe analysis of the input image should be a result of an evaluation process that involves different criteria like:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Accuracy: How well the image adjusts to the given description and the reference images.\n - Realism and coherence of the photo (the background makes sense, no absurd elements, no elements with weird shapes, etc.)\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be similar to the reference images. We don't want too much food.\n - Correct ingredients: The image should show the correct ingredients for the menu item.\n\n### MENU ITEM NAME:\nHouse Salad.\n\n### MENU ITEM DESCRIPTION:\nLettuce, tomatoes, onions, cilantro and house dressing..\n\nConsidering everything above, analyze and review the input image.\nI am going to give you a list of four issues that you connsider are preventing the image to pass a QA process.\nSelect the issue that best applies to the input image.\n\nThis is the list of possible issues: [oversaturated, ingredients-type, too-sharp, overexposed]\n\nReturn only the text from the issue that best applies to the input image.\n ", "\nI am going to give you an image of a menu item, which may be a food dish, a side or a beverage.\n\nThis input image was generated using AI. I am going to give you a list of four possible description for this menu item, separated by \" | \".\nSelect the one that best describes the image.\n\nThis is the list of possible descriptions: [Lettuce, tomatoes, onions, cilantro and house dressing. | Crispy noodle, peanuts crispy white meat chicken, lettuce, tomato, and cucumber. | Deep-fried chicken | Orange zesty brown sauce]\n\nReturn only the text from the list that best describes the menu item.\n ", "\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, and the images below, I need you to review the image and try to find any possible aesthetics-related issues with it.\n\nPlease consider the following criteria:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be reasonable. We don't want too much or too little food.\n\n### MENU ITEM NAME:\nHouse Salad.\n\n### MENU ITEM DESCRIPTION:\nLettuce, tomatoes, onions, cilantro and house dressing..\n\nConsidering everything above, analyze and review the input image.\nRespond only stating if the input image has any possible aesthetic issues or not.\n\nOnly possible responses are \"The image is good. It contains no significant aesthetic-related issues\" or \"The image has significant aesthetic issues that should be addressed\".\n" ]

[ "fail", "ingredients-type, ingredients-shape", "From the list of possible issues, the one that best applies to the input image is ingredients-type.", "The text that best describes the menu item is \"Lettuce, tomatoes, onions, cilantro and house dressing.\".", "The image is good. It contains no significant aesthetic-related issues." ]

[ "ingredients-type", "ingredients-shape" ]

[]

[]

[ "\n### CONTEXT\n\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n4. A set of one or more reference images of the same menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, I need you to review the image and assign a category to it.\n\nThe review result can only be one of three options: `fail`, `soft_fail`, or `pass`.\n\nThe review assigned to the input image should be a result of an evaluation process that involves different criteria like:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Accuracy: How well the image adjusts to the given description and the reference images.\n - Realism and coherence of the photo (the background makes sense, no absurd elements, no elements with weird shapes, etc.)\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be similar to the reference images. We don't want too much food.\n - Correct ingredients: The image should show the correct ingredients for the menu item.\n\n### MENU ITEM NAME:\nRice Pudding (Handmade).\n\n### MENU ITEM DESCRIPTION:\nBaked rice pudding with milt, rice and sugar. Serving with toppings..\n\nConsidering everything above, analyze and review the input image. I need your answer to be just the category that you think corresponds to the\nimage you will reviewing (fail, soft_fail, or pass). Don't return anything else.\n\nI repeat, ONLY return the category.\n ", "\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n4. A set of one or more reference images of the same menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, I need you to review the image and tray to find any possible issues with it.\n\nThe analysis of the input image should be a result of an evaluation process that involves different criteria like:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Accuracy: How well the image adjusts to the given description and the reference images.\n - Realism and coherence of the photo (the background makes sense, no absurd elements, no elements with weird shapes, etc.)\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be similar to the reference images. We don't want too much food.\n - Correct ingredients: The image should show the correct ingredients for the menu item.\n\n### MENU ITEM NAME:\nRice Pudding (Handmade).\n\n### MENU ITEM DESCRIPTION:\nBaked rice pudding with milt, rice and sugar. Serving with toppings..\n\nConsidering everything above, analyze and review the input image.\nI need your answer to be a list of one or more issues that you connsider are preventing the image to pass a QA process.\nIf you consider that the image has no significant issues and is ok to be approved, return \"No issues\".\n ", "\n### CONTEXT\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n4. A set of one or more reference images of the same menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, and the images below, I need you to review the image and try to find any possible issues with it.\n\nThe analysis of the input image should be a result of an evaluation process that involves different criteria like:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Accuracy: How well the image adjusts to the given description and the reference images.\n - Realism and coherence of the photo (the background makes sense, no absurd elements, no elements with weird shapes, etc.)\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be similar to the reference images. We don't want too much food.\n - Correct ingredients: The image should show the correct ingredients for the menu item.\n\n### MENU ITEM NAME:\nRice Pudding (Handmade).\n\n### MENU ITEM DESCRIPTION:\nBaked rice pudding with milt, rice and sugar. Serving with toppings..\n\nConsidering everything above, analyze and review the input image.\nI am going to give you a list of four issues that you connsider are preventing the image to pass a QA process.\nSelect the issue that best applies to the input image.\n\nThis is the list of possible issues: [ingredients-texture, highlights-too-bright, realistic-plating-arrangement, oversharpened]\n\nReturn only the text from the issue that best applies to the input image.\n ", "\nI am going to give you an image of a menu item, which may be a food dish, a side or a beverage.\n\nThis input image was generated using AI. I am going to give you a list of four possible description for this menu item, separated by \" | \".\nSelect the one that best describes the image.\n\nThis is the list of possible descriptions: [Baked rice pudding with milt, rice and sugar. Serving with toppings. | Lettuce, tomatoes, onions, cilantro and house dressing. | black tea, condensed milk, tapioca pearls]\n\nReturn only the text from the list that best describes the menu item.\n ", "\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, and the images below, I need you to review the image and try to find any possible aesthetics-related issues with it.\n\nPlease consider the following criteria:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be reasonable. We don't want too much or too little food.\n\n### MENU ITEM NAME:\nRice Pudding (Handmade).\n\n### MENU ITEM DESCRIPTION:\nBaked rice pudding with milt, rice and sugar. Serving with toppings..\n\nConsidering everything above, analyze and review the input image.\nRespond only stating if the input image has any possible aesthetic issues or not.\n\nOnly possible responses are \"The image is good. It contains no significant aesthetic-related issues\" or \"The image has significant aesthetic issues that should be addressed\".\n" ]

[ "soft_fail", "ingredients-texture", "From the list of possible issues, the one that best applies to the input image is ingredients-texture.", "The text that best describes the menu item is \"Baked rice pudding with milt, rice and sugar. Serving with toppings.\".", "The image is good. It contains no significant aesthetic-related issues." ]

[ "ingredients-texture" ]

[]

[]

[ "\n### CONTEXT\n\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n4. A set of one or more reference images of the same menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, I need you to review the image and assign a category to it.\n\nThe review result can only be one of three options: `fail`, `soft_fail`, or `pass`.\n\nThe review assigned to the input image should be a result of an evaluation process that involves different criteria like:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Accuracy: How well the image adjusts to the given description and the reference images.\n - Realism and coherence of the photo (the background makes sense, no absurd elements, no elements with weird shapes, etc.)\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be similar to the reference images. We don't want too much food.\n - Correct ingredients: The image should show the correct ingredients for the menu item.\n\n### MENU ITEM NAME:\nBreakfast, Ham, Egg, and Cheese.\n\n### MENU ITEM DESCRIPTION:\n<No description provided.>.\n\nConsidering everything above, analyze and review the input image. I need your answer to be just the category that you think corresponds to the\nimage you will reviewing (fail, soft_fail, or pass). Don't return anything else.\n\nI repeat, ONLY return the category.\n ", "\n### CONTEXT\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n4. A set of one or more reference images of the same menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, and the images below, I need you to review the image and try to find any possible issues with it.\n\nThe analysis of the input image should be a result of an evaluation process that involves different criteria like:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Accuracy: How well the image adjusts to the given description and the reference images.\n - Realism and coherence of the photo (the background makes sense, no absurd elements, no elements with weird shapes, etc.)\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be similar to the reference images. We don't want too much food.\n - Correct ingredients: The image should show the correct ingredients for the menu item.\n\n### MENU ITEM NAME:\nBreakfast, Ham, Egg, and Cheese.\n\n### MENU ITEM DESCRIPTION:\n<No description provided.>.\n\nConsidering everything above, analyze and review the input image.\nI am going to give you a list of four issues that you connsider are preventing the image to pass a QA process.\nSelect the issue that best applies to the input image.\n\nThis is the list of possible issues: [other, portion-amount, quantity-units, shadows-too-small]\n\nReturn only the text from the issue that best applies to the input image.\n ", "\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, and the images below, I need you to review the image and try to find any possible aesthetics-related issues with it.\n\nPlease consider the following criteria:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be reasonable. We don't want too much or too little food.\n\n### MENU ITEM NAME:\nBreakfast, Ham, Egg, and Cheese.\n\n### MENU ITEM DESCRIPTION:\n<No description provided.>.\n\nConsidering everything above, analyze and review the input image.\nRespond only stating if the input image has any possible aesthetic issues or not.\n\nOnly possible responses are \"The image is good. It contains no significant aesthetic-related issues\" or \"The image has significant aesthetic issues that should be addressed\".\n" ]

[ "fail", "From the list of possible issues, the one that best applies to the input image is other.", "The image is good. It contains no significant aesthetic-related issues." ]

[ "other" ]

[]

[]

[ "\n### CONTEXT\n\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n4. A set of one or more reference images of the same menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, I need you to review the image and assign a category to it.\n\nThe review result can only be one of three options: `fail`, `soft_fail`, or `pass`.\n\nThe review assigned to the input image should be a result of an evaluation process that involves different criteria like:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Accuracy: How well the image adjusts to the given description and the reference images.\n - Realism and coherence of the photo (the background makes sense, no absurd elements, no elements with weird shapes, etc.)\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be similar to the reference images. We don't want too much food.\n - Correct ingredients: The image should show the correct ingredients for the menu item.\n\n### MENU ITEM NAME:\nKeto Beyond Salad (V).\n\n### MENU ITEM DESCRIPTION:\nMixed green salad served with seasoned Beyond (plant-based) meatballs, grape tomato, cucumber, avocado, pickled red onion, parsley and lemon-mustard vinaigrette. Gluten-free. Vegan..\n\nConsidering everything above, analyze and review the input image. I need your answer to be just the category that you think corresponds to the\nimage you will reviewing (fail, soft_fail, or pass). Don't return anything else.\n\nI repeat, ONLY return the category.\n ", "\n### CONTEXT\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n4. A set of one or more reference images of the same menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, and the images below, I need you to review the image and try to find any possible issues with it.\n\nThe analysis of the input image should be a result of an evaluation process that involves different criteria like:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Accuracy: How well the image adjusts to the given description and the reference images.\n - Realism and coherence of the photo (the background makes sense, no absurd elements, no elements with weird shapes, etc.)\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be similar to the reference images. We don't want too much food.\n - Correct ingredients: The image should show the correct ingredients for the menu item.\n\n### MENU ITEM NAME:\nKeto Beyond Salad (V).\n\n### MENU ITEM DESCRIPTION:\nMixed green salad served with seasoned Beyond (plant-based) meatballs, grape tomato, cucumber, avocado, pickled red onion, parsley and lemon-mustard vinaigrette. Gluten-free. Vegan..\n\nConsidering everything above, analyze and review the input image.\nI am going to give you a list of four issues that you connsider are preventing the image to pass a QA process.\nSelect the issue that best applies to the input image.\n\nThis is the list of possible issues: [portion-of-image-is-too-similar-to-other-parts, shadow-too-tight, ingredients-size, other]\n\nReturn only the text from the issue that best applies to the input image.\n ", "\nI am going to give you an image of a menu item, which may be a food dish, a side or a beverage.\n\nThis input image was generated using AI. I am going to give you a list of four possible description for this menu item, separated by \" | \".\nSelect the one that best describes the image.\n\nThis is the list of possible descriptions: [Mixed green salad served with seasoned Beyond (plant-based) meatballs, grape tomato, cucumber, avocado, pickled red onion, parsley and lemon-mustard vinaigrette. Gluten-free. Vegan. | Seasoned broth with filled wonton dumplings with beef stew. Savory light broth with noodles. | Hot pastrami, turkey, cheese, corned beef, and jalapeños.]\n\nReturn only the text from the list that best describes the menu item.\n ", "\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, and the images below, I need you to review the image and try to find any possible aesthetics-related issues with it.\n\nPlease consider the following criteria:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be reasonable. We don't want too much or too little food.\n\n### MENU ITEM NAME:\nKeto Beyond Salad (V).\n\n### MENU ITEM DESCRIPTION:\nMixed green salad served with seasoned Beyond (plant-based) meatballs, grape tomato, cucumber, avocado, pickled red onion, parsley and lemon-mustard vinaigrette. Gluten-free. Vegan..\n\nConsidering everything above, analyze and review the input image.\nRespond only stating if the input image has any possible aesthetic issues or not.\n\nOnly possible responses are \"The image is good. It contains no significant aesthetic-related issues\" or \"The image has significant aesthetic issues that should be addressed\".\n" ]

[ "soft_fail", "From the list of possible issues, the one that best applies to the input image is other.", "The text that best describes the menu item is \"Mixed green salad served with seasoned Beyond (plant-based) meatballs, grape tomato, cucumber, avocado, pickled red onion, parsley and lemon-mustard vinaigrette. Gluten-free. Vegan.\".", "The image has significant aesthetic issues that should be addressed." ]

[]

[ "other" ]

[]

[ "\n### CONTEXT\n\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n4. A set of one or more reference images of the same menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, I need you to review the image and assign a category to it.\n\nThe review result can only be one of three options: `fail`, `soft_fail`, or `pass`.\n\nThe review assigned to the input image should be a result of an evaluation process that involves different criteria like:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Accuracy: How well the image adjusts to the given description and the reference images.\n - Realism and coherence of the photo (the background makes sense, no absurd elements, no elements with weird shapes, etc.)\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be similar to the reference images. We don't want too much food.\n - Correct ingredients: The image should show the correct ingredients for the menu item.\n\n### MENU ITEM NAME:\nThe Big Knockout.\n\n### MENU ITEM DESCRIPTION:\ninclude: butter, cheese, sour cream, green onions and bacon bits.\n\nConsidering everything above, analyze and review the input image. I need your answer to be just the category that you think corresponds to the\nimage you will reviewing (fail, soft_fail, or pass). Don't return anything else.\n\nI repeat, ONLY return the category.\n ", "\n### CONTEXT\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n4. A set of one or more reference images of the same menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, and the images below, I need you to review the image and try to find any possible issues with it.\n\nThe analysis of the input image should be a result of an evaluation process that involves different criteria like:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Accuracy: How well the image adjusts to the given description and the reference images.\n - Realism and coherence of the photo (the background makes sense, no absurd elements, no elements with weird shapes, etc.)\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be similar to the reference images. We don't want too much food.\n - Correct ingredients: The image should show the correct ingredients for the menu item.\n\n### MENU ITEM NAME:\nThe Big Knockout.\n\n### MENU ITEM DESCRIPTION:\ninclude: butter, cheese, sour cream, green onions and bacon bits.\n\nConsidering everything above, analyze and review the input image.\nI am going to give you a list of four issues that you connsider are preventing the image to pass a QA process.\nSelect the issue that best applies to the input image.\n\nThis is the list of possible issues: [too-sharp, shadow-too-dispersed, incorrect-perspective, other]\n\nReturn only the text from the issue that best applies to the input image.\n ", "\nI am going to give you an image of a menu item, which may be a food dish, a side or a beverage.\n\nThis input image was generated using AI. I am going to give you a list of four possible description for this menu item, separated by \" | \".\nSelect the one that best describes the image.\n\nThis is the list of possible descriptions: [include: butter, cheese, sour cream, green onions and bacon bits | One of our most popular cakes. Moist sponge layers brushed with an orange simple syrup in between is a thin layer of raspberry preserves, Italian pastry cream, and fresh whipped cream. This cake is then covered in almond paste. | Includes: mustard, mayo, lettuce, tomato, pickles, and onions]\n\nReturn only the text from the list that best describes the menu item.\n ", "\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, and the images below, I need you to review the image and try to find any possible aesthetics-related issues with it.\n\nPlease consider the following criteria:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be reasonable. We don't want too much or too little food.\n\n### MENU ITEM NAME:\nThe Big Knockout.\n\n### MENU ITEM DESCRIPTION:\ninclude: butter, cheese, sour cream, green onions and bacon bits.\n\nConsidering everything above, analyze and review the input image.\nRespond only stating if the input image has any possible aesthetic issues or not.\n\nOnly possible responses are \"The image is good. It contains no significant aesthetic-related issues\" or \"The image has significant aesthetic issues that should be addressed\".\n" ]

[ "soft_fail", "From the list of possible issues, the one that best applies to the input image is other.", "The text that best describes the menu item is \"include: butter, cheese, sour cream, green onions and bacon bits\".", "The image has significant aesthetic issues that should be addressed." ]

[]

[ "other" ]

[]

[ "\n### CONTEXT\n\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n4. A set of one or more reference images of the same menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, I need you to review the image and assign a category to it.\n\nThe review result can only be one of three options: `fail`, `soft_fail`, or `pass`.\n\nThe review assigned to the input image should be a result of an evaluation process that involves different criteria like:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Accuracy: How well the image adjusts to the given description and the reference images.\n - Realism and coherence of the photo (the background makes sense, no absurd elements, no elements with weird shapes, etc.)\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be similar to the reference images. We don't want too much food.\n - Correct ingredients: The image should show the correct ingredients for the menu item.\n\n### MENU ITEM NAME:\nCajun Pizza (18\").\n\n### MENU ITEM DESCRIPTION:\nChicken breast marinated with cajun spice, tomatoes, bell peppers, onions, garlic, mozzarella cheese, and cajun sauce..\n\nConsidering everything above, analyze and review the input image. I need your answer to be just the category that you think corresponds to the\nimage you will reviewing (fail, soft_fail, or pass). Don't return anything else.\n\nI repeat, ONLY return the category.\n ", "\n### CONTEXT\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n4. A set of one or more reference images of the same menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, and the images below, I need you to review the image and try to find any possible issues with it.\n\nThe analysis of the input image should be a result of an evaluation process that involves different criteria like:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Accuracy: How well the image adjusts to the given description and the reference images.\n - Realism and coherence of the photo (the background makes sense, no absurd elements, no elements with weird shapes, etc.)\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be similar to the reference images. We don't want too much food.\n - Correct ingredients: The image should show the correct ingredients for the menu item.\n\n### MENU ITEM NAME:\nCajun Pizza (18\").\n\n### MENU ITEM DESCRIPTION:\nChicken breast marinated with cajun spice, tomatoes, bell peppers, onions, garlic, mozzarella cheese, and cajun sauce..\n\nConsidering everything above, analyze and review the input image.\nI am going to give you a list of four issues that you connsider are preventing the image to pass a QA process.\nSelect the issue that best applies to the input image.\n\nThis is the list of possible issues: [uneven-spread-of-light, other, food-arrangement-too-similar-to-reference-image, container-too-big]\n\nReturn only the text from the issue that best applies to the input image.\n ", "\nI am going to give you an image of a menu item, which may be a food dish, a side or a beverage.\n\nThis input image was generated using AI. I am going to give you a list of four possible description for this menu item, separated by \" | \".\nSelect the one that best describes the image.\n\nThis is the list of possible descriptions: [Red beans slowly cooked with ham hocks served with rice, collard greens, and grits. | Chicken breast marinated with cajun spice, tomatoes, bell peppers, onions, garlic, mozzarella cheese, and cajun sauce. | 20 pieces of corn nuggets]\n\nReturn only the text from the list that best describes the menu item.\n ", "\nI am going to give you the following inputs:\n\n1. An image of a menu item, which may be a food dish, a side or a beverage. This will be referred to as the \"input image\".\n2. The name of the menu item\n3. The description of the menu item.\n\nThe input image is generated using AI, and is intented to be shown to customers in digital menus.\nThe input image should match the description of the menu item and look similar to the reference images, while looking aesthetically pleasing.\n\nBased on the given inputs, and the images below, I need you to review the image and try to find any possible aesthetics-related issues with it.\n\nPlease consider the following criteria:\n - Aesthetics: the input image should look good and the food should be appetizing.\n - Overall quality and composition of the photo (lighting, shadows, resoultion, etc.)\n - Size: The size (portion) of the menu item should be reasonable. We don't want too much or too little food.\n\n### MENU ITEM NAME:\nCajun Pizza (18\").\n\n### MENU ITEM DESCRIPTION:\nChicken breast marinated with cajun spice, tomatoes, bell peppers, onions, garlic, mozzarella cheese, and cajun sauce..\n\nConsidering everything above, analyze and review the input image.\nRespond only stating if the input image has any possible aesthetic issues or not.\n\nOnly possible responses are \"The image is good. It contains no significant aesthetic-related issues\" or \"The image has significant aesthetic issues that should be addressed\".\n" ]

[ "soft_fail", "From the list of possible issues, the one that best applies to the input image is other.", "The text that best describes the menu item is \"Chicken breast marinated with cajun spice, tomatoes, bell peppers, onions, garlic, mozzarella cheese, and cajun sauce.\".", "The image has significant aesthetic issues that should be addressed." ]

[]

[ "other" ]

[]