FaceAttribNet: Optimized for Mobile Deployment

Comprehensive facial analysis by extracting face features

Facial feature extraction and additional attributes including liveness, eyeclose, mask and glasses detection for face recognition.

This model is an implementation of FaceAttribNet found here.

This repository provides scripts to run FaceAttribNet on Qualcomm® devices. More details on model performance across various devices, can be found here.

Model Details

Model Type: Object detection
Model Stats:
- Model checkpoint: multitask_FR_state_dict.pt
- Input resolution: 128x128
- Number of parameters: 11.6M
- Model size: 47.6MB

Model	Device	Chipset	Target Runtime	Inference Time (ms)	Peak Memory Range (MB)	Precision	Primary Compute Unit	Target Model
FaceAttribNet	Samsung Galaxy S23	Snapdragon® 8 Gen 2	TFLITE	0.957 ms	0 - 2 MB	FP16	NPU	FaceAttribNet.tflite
FaceAttribNet	Samsung Galaxy S23	Snapdragon® 8 Gen 2	QNN	1.041 ms	0 - 8 MB	FP16	NPU	FaceAttribNet.so
FaceAttribNet	Samsung Galaxy S23	Snapdragon® 8 Gen 2	ONNX	1.2 ms	0 - 364 MB	FP16	NPU	FaceAttribNet.onnx
FaceAttribNet	Samsung Galaxy S24	Snapdragon® 8 Gen 3	TFLITE	0.71 ms	0 - 110 MB	FP16	NPU	FaceAttribNet.tflite
FaceAttribNet	Samsung Galaxy S24	Snapdragon® 8 Gen 3	QNN	0.781 ms	0 - 20 MB	FP16	NPU	FaceAttribNet.so
FaceAttribNet	Samsung Galaxy S24	Snapdragon® 8 Gen 3	ONNX	0.907 ms	0 - 113 MB	FP16	NPU	FaceAttribNet.onnx
FaceAttribNet	Snapdragon 8 Elite QRD	Snapdragon® 8 Elite	TFLITE	0.658 ms	0 - 60 MB	FP16	NPU	FaceAttribNet.tflite
FaceAttribNet	Snapdragon 8 Elite QRD	Snapdragon® 8 Elite	QNN	0.702 ms	0 - 21 MB	FP16	NPU	Use Export Script
FaceAttribNet	Snapdragon 8 Elite QRD	Snapdragon® 8 Elite	ONNX	0.76 ms	0 - 61 MB	FP16	NPU	FaceAttribNet.onnx
FaceAttribNet	QCS8550 (Proxy)	QCS8550 Proxy	TFLITE	0.947 ms	0 - 1 MB	FP16	NPU	FaceAttribNet.tflite
FaceAttribNet	QCS8550 (Proxy)	QCS8550 Proxy	QNN	0.998 ms	0 - 1 MB	FP16	NPU	Use Export Script
FaceAttribNet	SA8255 (Proxy)	SA8255P Proxy	TFLITE	0.977 ms	0 - 2 MB	FP16	NPU	FaceAttribNet.tflite
FaceAttribNet	SA8255 (Proxy)	SA8255P Proxy	QNN	1.001 ms	0 - 1 MB	FP16	NPU	Use Export Script
FaceAttribNet	SA8775 (Proxy)	SA8775P Proxy	TFLITE	0.978 ms	0 - 2 MB	FP16	NPU	FaceAttribNet.tflite
FaceAttribNet	SA8775 (Proxy)	SA8775P Proxy	QNN	0.998 ms	0 - 1 MB	FP16	NPU	Use Export Script
FaceAttribNet	SA8650 (Proxy)	SA8650P Proxy	TFLITE	0.987 ms	0 - 2 MB	FP16	NPU	FaceAttribNet.tflite
FaceAttribNet	SA8650 (Proxy)	SA8650P Proxy	QNN	0.996 ms	0 - 2 MB	FP16	NPU	Use Export Script
FaceAttribNet	SA8295P ADP	SA8295P	TFLITE	1.663 ms	0 - 52 MB	FP16	NPU	FaceAttribNet.tflite
FaceAttribNet	SA8295P ADP	SA8295P	QNN	1.742 ms	0 - 6 MB	FP16	NPU	Use Export Script
FaceAttribNet	QCS8450 (Proxy)	QCS8450 Proxy	TFLITE	1.258 ms	0 - 102 MB	FP16	NPU	FaceAttribNet.tflite
FaceAttribNet	QCS8450 (Proxy)	QCS8450 Proxy	QNN	1.356 ms	0 - 21 MB	FP16	NPU	Use Export Script
FaceAttribNet	Snapdragon X Elite CRD	Snapdragon® X Elite	QNN	1.14 ms	0 - 0 MB	FP16	NPU	Use Export Script
FaceAttribNet	Snapdragon X Elite CRD	Snapdragon® X Elite	ONNX	1.213 ms	27 - 27 MB	FP16	NPU	FaceAttribNet.onnx

Installation

This model can be installed as a Python package via pip.

pip install qai-hub-models

Configure Qualcomm® AI Hub to run this model on a cloud-hosted device

Sign-in to Qualcomm® AI Hub with your Qualcomm® ID. Once signed in navigate to Account -> Settings -> API Token.

With this API token, you can configure your client to run models on the cloud hosted devices.

qai-hub configure --api_token API_TOKEN

Navigate to docs for more information.

Demo off target

The package contains a simple end-to-end demo that downloads pre-trained weights and runs this model on a sample input.

python -m qai_hub_models.models.face_attrib_net.demo

The above demo runs a reference implementation of pre-processing, model inference, and post processing.

NOTE: If you want running in a Jupyter Notebook or Google Colab like environment, please add the following to your cell (instead of the above).

%run -m qai_hub_models.models.face_attrib_net.demo

Run model on a cloud-hosted device

In addition to the demo, you can also run the model on a cloud-hosted Qualcomm® device. This script does the following:

Performance check on-device on a cloud-hosted device
Downloads compiled assets that can be deployed on-device for Android.
Accuracy check between PyTorch and on-device outputs.

python -m qai_hub_models.models.face_attrib_net.export

Profiling Results
------------------------------------------------------------
FaceAttribNet
Device                          : Samsung Galaxy S23 (13)
Runtime                         : TFLITE                 
Estimated inference time (ms)   : 1.0                    
Estimated peak memory usage (MB): [0, 2]                 
Total # Ops                     : 161                    
Compute Unit(s)                 : NPU (161 ops)

How does this work?

This export script leverages Qualcomm® AI Hub to optimize, validate, and deploy this model on-device. Lets go through each step below in detail:

Step 1: Compile model for on-device deployment

To compile a PyTorch model for on-device deployment, we first trace the model in memory using the jit.trace and then call the submit_compile_job API.

import torch

import qai_hub as hub
from qai_hub_models.models.face_attrib_net import 

# Load the model

# Device
device = hub.Device("Samsung Galaxy S23")

Step 2: Performance profiling on cloud-hosted device

After compiling models from step 1. Models can be profiled model on-device using the target_model. Note that this scripts runs the model on a device automatically provisioned in the cloud. Once the job is submitted, you can navigate to a provided job URL to view a variety of on-device performance metrics.

profile_job = hub.submit_profile_job(
    model=target_model,
    device=device,
)

Step 3: Verify on-device accuracy

To verify the accuracy of the model on-device, you can run on-device inference on sample input data on the same cloud hosted device.

input_data = torch_model.sample_inputs()
inference_job = hub.submit_inference_job(
    model=target_model,
    device=device,
    inputs=input_data,
)
    on_device_output = inference_job.download_output_data()

With the output of the model, you can compute like PSNR, relative errors or spot check the output with expected output.

Note: This on-device profiling and inference requires access to Qualcomm® AI Hub. Sign up for access.

Deploying compiled model to Android

The models can be deployed using multiple runtimes:

TensorFlow Lite (.tflite export): This tutorial provides a guide to deploy the .tflite model in an Android application.
QNN (.so export ): This sample app provides instructions on how to use the .so shared library in an Android application.

View on Qualcomm® AI Hub

Get more details on FaceAttribNet's performance across various devices here. Explore all available models on Qualcomm® AI Hub

License

The license for the original implementation of FaceAttribNet can be found here.
The license for the compiled assets for on-device deployment can be found here

References

Community

Join our AI Hub Slack community to collaborate, post questions and learn more about on-device AI.
For questions or feedback please reach out to us.