Spaces:
Running
Running
- __init_paths.py +3 -0
- app.py +4 -1
- face_enhancement.py +17 -2
- sr_model/arch_util.py +125 -0
- sr_model/real_esrnet.py +133 -0
- sr_model/rrdbnet_arch.py +116 -0
__init_paths.py
CHANGED
@@ -14,5 +14,8 @@ this_dir = osp.dirname(__file__)
|
|
14 |
path = osp.join(this_dir, 'retinaface')
|
15 |
add_path(path)
|
16 |
|
|
|
|
|
|
|
17 |
path = osp.join(this_dir, 'face_model')
|
18 |
add_path(path)
|
|
|
14 |
path = osp.join(this_dir, 'retinaface')
|
15 |
add_path(path)
|
16 |
|
17 |
+
path = osp.join(this_dir, 'sr_model')
|
18 |
+
add_path(path)
|
19 |
+
|
20 |
path = osp.join(this_dir, 'face_model')
|
21 |
add_path(path)
|
app.py
CHANGED
@@ -3,6 +3,7 @@ import os
|
|
3 |
os.system('wget "https://public-vigen-video.oss-cn-shanghai.aliyuncs.com/robin/models/RetinaFace-R50.pth?OSSAccessKeyId=LTAI4G6bfnyW4TA4wFUXTYBe&Expires=1961116085&Signature=GlUNW6%2B8FxvxWmE9jKIZYOOciKQ%3D" -O weights/RetinaFace-R50.pth')
|
4 |
os.system('wget "https://public-vigen-video.oss-cn-shanghai.aliyuncs.com/robin/models/GPEN-BFR-512.pth?OSSAccessKeyId=LTAI4G6bfnyW4TA4wFUXTYBe&Expires=1961116208&Signature=hBgvVvKVSNGeXqT8glG%2Bd2t2OKc%3D" -O weights/GPEN-512.pth')
|
5 |
os.system('wget "https://public-vigen-video.oss-cn-shanghai.aliyuncs.com/robin/models/GPEN-Colorization-1024.pth?OSSAccessKeyId=LTAI4G6bfnyW4TA4wFUXTYBe&Expires=1961116315&Signature=9tPavW2h%2F1LhIKiXj73sTQoWqcc%3D" -O weights/GPEN-1024-Color.pth ')
|
|
|
6 |
|
7 |
import gradio as gr
|
8 |
|
@@ -21,6 +22,7 @@ def inference(file, mode):
|
|
21 |
if mode == "enhance":
|
22 |
model = {'name':'GPEN-512', 'size':512}
|
23 |
im = cv2.imread(file, cv2.IMREAD_COLOR)
|
|
|
24 |
faceenhancer = FaceEnhancement(size=model['size'], model=model['name'], channel_multiplier=2, device='cpu')
|
25 |
img, orig_faces, enhanced_faces = faceenhancer.process(im)
|
26 |
cv2.imwrite(os.path.join("output.png"), img)
|
@@ -50,7 +52,8 @@ gr.Interface(
|
|
50 |
description=description,
|
51 |
article=article,
|
52 |
examples=[
|
53 |
-
['
|
|
|
54 |
],
|
55 |
enable_queue=True
|
56 |
).launch()
|
|
|
3 |
os.system('wget "https://public-vigen-video.oss-cn-shanghai.aliyuncs.com/robin/models/RetinaFace-R50.pth?OSSAccessKeyId=LTAI4G6bfnyW4TA4wFUXTYBe&Expires=1961116085&Signature=GlUNW6%2B8FxvxWmE9jKIZYOOciKQ%3D" -O weights/RetinaFace-R50.pth')
|
4 |
os.system('wget "https://public-vigen-video.oss-cn-shanghai.aliyuncs.com/robin/models/GPEN-BFR-512.pth?OSSAccessKeyId=LTAI4G6bfnyW4TA4wFUXTYBe&Expires=1961116208&Signature=hBgvVvKVSNGeXqT8glG%2Bd2t2OKc%3D" -O weights/GPEN-512.pth')
|
5 |
os.system('wget "https://public-vigen-video.oss-cn-shanghai.aliyuncs.com/robin/models/GPEN-Colorization-1024.pth?OSSAccessKeyId=LTAI4G6bfnyW4TA4wFUXTYBe&Expires=1961116315&Signature=9tPavW2h%2F1LhIKiXj73sTQoWqcc%3D" -O weights/GPEN-1024-Color.pth ')
|
6 |
+
os.system('wget "https://public-vigen-video.oss-cn-shanghai.aliyuncs.com/robin/models/realesrnet_x2.pth?OSSAccessKeyId=LTAI4G6bfnyW4TA4wFUXTYBe&Expires=1962694780&Signature=lI%2FolhA%2FyigiTRvoDIVbtMIyhjI%3D" -O weights/realesrnet_x2.pth ')
|
7 |
|
8 |
import gradio as gr
|
9 |
|
|
|
22 |
if mode == "enhance":
|
23 |
model = {'name':'GPEN-512', 'size':512}
|
24 |
im = cv2.imread(file, cv2.IMREAD_COLOR)
|
25 |
+
im = cv2.resize(im, (0,0), fx=2, fy=2)
|
26 |
faceenhancer = FaceEnhancement(size=model['size'], model=model['name'], channel_multiplier=2, device='cpu')
|
27 |
img, orig_faces, enhanced_faces = faceenhancer.process(im)
|
28 |
cv2.imwrite(os.path.join("output.png"), img)
|
|
|
52 |
description=description,
|
53 |
article=article,
|
54 |
examples=[
|
55 |
+
['enhance.png'],
|
56 |
+
['color.png']
|
57 |
],
|
58 |
enable_queue=True
|
59 |
).launch()
|
face_enhancement.py
CHANGED
@@ -11,12 +11,14 @@ from PIL import Image
|
|
11 |
import __init_paths
|
12 |
from retinaface.retinaface_detection import RetinaFaceDetection
|
13 |
from face_model.face_gan import FaceGAN
|
|
|
14 |
from align_faces import warp_and_crop_face, get_reference_facial_points
|
15 |
|
16 |
class FaceEnhancement(object):
|
17 |
-
def __init__(self, base_dir='./', size=512, out_size=None, model=None, channel_multiplier=2, narrow=1, key=None, device='
|
18 |
self.facedetector = RetinaFaceDetection(base_dir, device)
|
19 |
self.facegan = FaceGAN(base_dir, size, out_size, model, channel_multiplier, narrow, key, device=device)
|
|
|
20 |
self.size = size
|
21 |
self.out_size = size if out_size==None else out_size
|
22 |
self.threshold = 0.9
|
@@ -53,6 +55,16 @@ class FaceEnhancement(object):
|
|
53 |
orig_faces.append(img)
|
54 |
enhanced_faces.append(ef)
|
55 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
56 |
facebs, landms = self.facedetector.detect(img)
|
57 |
|
58 |
height, width = img.shape[:2]
|
@@ -89,7 +101,10 @@ class FaceEnhancement(object):
|
|
89 |
full_img[np.where(mask>0)] = tmp_img[np.where(mask>0)]
|
90 |
|
91 |
full_mask = full_mask[:, :, np.newaxis]
|
92 |
-
|
|
|
|
|
|
|
93 |
|
94 |
return img, orig_faces, enhanced_faces
|
95 |
|
|
|
11 |
import __init_paths
|
12 |
from retinaface.retinaface_detection import RetinaFaceDetection
|
13 |
from face_model.face_gan import FaceGAN
|
14 |
+
from sr_model.real_esrnet import RealESRNet
|
15 |
from align_faces import warp_and_crop_face, get_reference_facial_points
|
16 |
|
17 |
class FaceEnhancement(object):
|
18 |
+
def __init__(self, base_dir='./', size=512, out_size=None, model=None, channel_multiplier=2, narrow=1, key=None, device='cpu'):
|
19 |
self.facedetector = RetinaFaceDetection(base_dir, device)
|
20 |
self.facegan = FaceGAN(base_dir, size, out_size, model, channel_multiplier, narrow, key, device=device)
|
21 |
+
self.srmodel = RealESRNet(base_dir, args.sr_model, args.sr_scale, args.tile_size, device=device)
|
22 |
self.size = size
|
23 |
self.out_size = size if out_size==None else out_size
|
24 |
self.threshold = 0.9
|
|
|
55 |
orig_faces.append(img)
|
56 |
enhanced_faces.append(ef)
|
57 |
|
58 |
+
if self.use_sr:
|
59 |
+
ef = self.srmodel.process(ef)
|
60 |
+
|
61 |
+
return ef, orig_faces, enhanced_faces
|
62 |
+
|
63 |
+
if self.use_sr:
|
64 |
+
img_sr = self.srmodel.process(img)
|
65 |
+
if img_sr is not None:
|
66 |
+
img = cv2.resize(img, img_sr.shape[:2][::-1])
|
67 |
+
|
68 |
facebs, landms = self.facedetector.detect(img)
|
69 |
|
70 |
height, width = img.shape[:2]
|
|
|
101 |
full_img[np.where(mask>0)] = tmp_img[np.where(mask>0)]
|
102 |
|
103 |
full_mask = full_mask[:, :, np.newaxis]
|
104 |
+
if self.use_sr and img_sr is not None:
|
105 |
+
img = cv2.convertScaleAbs(img_sr*(1-full_mask) + full_img*full_mask)
|
106 |
+
else:
|
107 |
+
img = cv2.convertScaleAbs(img*(1-full_mask) + full_img*full_mask)
|
108 |
|
109 |
return img, orig_faces, enhanced_faces
|
110 |
|
sr_model/arch_util.py
ADDED
@@ -0,0 +1,125 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import math
|
2 |
+
import torch
|
3 |
+
from torch import nn as nn
|
4 |
+
from torch.nn import functional as F
|
5 |
+
from torch.nn import init as init
|
6 |
+
from torch.nn.modules.batchnorm import _BatchNorm
|
7 |
+
|
8 |
+
@torch.no_grad()
|
9 |
+
def default_init_weights(module_list, scale=1, bias_fill=0, **kwargs):
|
10 |
+
"""Initialize network weights.
|
11 |
+
|
12 |
+
Args:
|
13 |
+
module_list (list[nn.Module] | nn.Module): Modules to be initialized.
|
14 |
+
scale (float): Scale initialized weights, especially for residual
|
15 |
+
blocks. Default: 1.
|
16 |
+
bias_fill (float): The value to fill bias. Default: 0
|
17 |
+
kwargs (dict): Other arguments for initialization function.
|
18 |
+
"""
|
19 |
+
if not isinstance(module_list, list):
|
20 |
+
module_list = [module_list]
|
21 |
+
for module in module_list:
|
22 |
+
for m in module.modules():
|
23 |
+
if isinstance(m, nn.Conv2d):
|
24 |
+
init.kaiming_normal_(m.weight, **kwargs)
|
25 |
+
m.weight.data *= scale
|
26 |
+
if m.bias is not None:
|
27 |
+
m.bias.data.fill_(bias_fill)
|
28 |
+
elif isinstance(m, nn.Linear):
|
29 |
+
init.kaiming_normal_(m.weight, **kwargs)
|
30 |
+
m.weight.data *= scale
|
31 |
+
if m.bias is not None:
|
32 |
+
m.bias.data.fill_(bias_fill)
|
33 |
+
elif isinstance(m, _BatchNorm):
|
34 |
+
init.constant_(m.weight, 1)
|
35 |
+
if m.bias is not None:
|
36 |
+
m.bias.data.fill_(bias_fill)
|
37 |
+
|
38 |
+
|
39 |
+
def make_layer(basic_block, num_basic_block, **kwarg):
|
40 |
+
"""Make layers by stacking the same blocks.
|
41 |
+
|
42 |
+
Args:
|
43 |
+
basic_block (nn.module): nn.module class for basic block.
|
44 |
+
num_basic_block (int): number of blocks.
|
45 |
+
|
46 |
+
Returns:
|
47 |
+
nn.Sequential: Stacked blocks in nn.Sequential.
|
48 |
+
"""
|
49 |
+
layers = []
|
50 |
+
for _ in range(num_basic_block):
|
51 |
+
layers.append(basic_block(**kwarg))
|
52 |
+
return nn.Sequential(*layers)
|
53 |
+
|
54 |
+
|
55 |
+
class ResidualBlockNoBN(nn.Module):
|
56 |
+
"""Residual block without BN.
|
57 |
+
|
58 |
+
It has a style of:
|
59 |
+
---Conv-ReLU-Conv-+-
|
60 |
+
|________________|
|
61 |
+
|
62 |
+
Args:
|
63 |
+
num_feat (int): Channel number of intermediate features.
|
64 |
+
Default: 64.
|
65 |
+
res_scale (float): Residual scale. Default: 1.
|
66 |
+
pytorch_init (bool): If set to True, use pytorch default init,
|
67 |
+
otherwise, use default_init_weights. Default: False.
|
68 |
+
"""
|
69 |
+
|
70 |
+
def __init__(self, num_feat=64, res_scale=1, pytorch_init=False):
|
71 |
+
super(ResidualBlockNoBN, self).__init__()
|
72 |
+
self.res_scale = res_scale
|
73 |
+
self.conv1 = nn.Conv2d(num_feat, num_feat, 3, 1, 1, bias=True)
|
74 |
+
self.conv2 = nn.Conv2d(num_feat, num_feat, 3, 1, 1, bias=True)
|
75 |
+
self.relu = nn.ReLU(inplace=True)
|
76 |
+
|
77 |
+
if not pytorch_init:
|
78 |
+
default_init_weights([self.conv1, self.conv2], 0.1)
|
79 |
+
|
80 |
+
def forward(self, x):
|
81 |
+
identity = x
|
82 |
+
out = self.conv2(self.relu(self.conv1(x)))
|
83 |
+
return identity + out * self.res_scale
|
84 |
+
|
85 |
+
|
86 |
+
class Upsample(nn.Sequential):
|
87 |
+
"""Upsample module.
|
88 |
+
|
89 |
+
Args:
|
90 |
+
scale (int): Scale factor. Supported scales: 2^n and 3.
|
91 |
+
num_feat (int): Channel number of intermediate features.
|
92 |
+
"""
|
93 |
+
|
94 |
+
def __init__(self, scale, num_feat):
|
95 |
+
m = []
|
96 |
+
if (scale & (scale - 1)) == 0: # scale = 2^n
|
97 |
+
for _ in range(int(math.log(scale, 2))):
|
98 |
+
m.append(nn.Conv2d(num_feat, 4 * num_feat, 3, 1, 1))
|
99 |
+
m.append(nn.PixelShuffle(2))
|
100 |
+
elif scale == 3:
|
101 |
+
m.append(nn.Conv2d(num_feat, 9 * num_feat, 3, 1, 1))
|
102 |
+
m.append(nn.PixelShuffle(3))
|
103 |
+
else:
|
104 |
+
raise ValueError(f'scale {scale} is not supported. '
|
105 |
+
'Supported scales: 2^n and 3.')
|
106 |
+
super(Upsample, self).__init__(*m)
|
107 |
+
|
108 |
+
# TODO: may write a cpp file
|
109 |
+
def pixel_unshuffle(x, scale):
|
110 |
+
""" Pixel unshuffle.
|
111 |
+
|
112 |
+
Args:
|
113 |
+
x (Tensor): Input feature with shape (b, c, hh, hw).
|
114 |
+
scale (int): Downsample ratio.
|
115 |
+
|
116 |
+
Returns:
|
117 |
+
Tensor: the pixel unshuffled feature.
|
118 |
+
"""
|
119 |
+
b, c, hh, hw = x.size()
|
120 |
+
out_channel = c * (scale**2)
|
121 |
+
assert hh % scale == 0 and hw % scale == 0
|
122 |
+
h = hh // scale
|
123 |
+
w = hw // scale
|
124 |
+
x_view = x.view(b, c, h, scale, w, scale)
|
125 |
+
return x_view.permute(0, 1, 3, 5, 2, 4).reshape(b, out_channel, h, w)
|
sr_model/real_esrnet.py
ADDED
@@ -0,0 +1,133 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import os
|
2 |
+
import math
|
3 |
+
import torch
|
4 |
+
import numpy as np
|
5 |
+
from rrdbnet_arch import RRDBNet
|
6 |
+
from torch.nn import functional as F
|
7 |
+
|
8 |
+
class RealESRNet(object):
|
9 |
+
def __init__(self, base_dir='./', model=None, scale=2, tile_size=0, tile_pad=10, device='cuda'):
|
10 |
+
self.base_dir = base_dir
|
11 |
+
self.scale = scale
|
12 |
+
self.tile_size = tile_size
|
13 |
+
self.tile_pad = tile_pad
|
14 |
+
self.device = device
|
15 |
+
self.load_srmodel(base_dir, model)
|
16 |
+
|
17 |
+
def load_srmodel(self, base_dir, model):
|
18 |
+
self.srmodel = RRDBNet(num_in_ch=3, num_out_ch=3, num_feat=32, num_block=23, num_grow_ch=32, scale=self.scale)
|
19 |
+
if model is None:
|
20 |
+
loadnet = torch.load(os.path.join(self.base_dir, 'weights', 'realesrnet_x%d.pth'%self.scale))
|
21 |
+
else:
|
22 |
+
loadnet = torch.load(os.path.join(self.base_dir, 'weights', model+'_x%d.pth'%self.scale))
|
23 |
+
#print(loadnet['params_ema'].keys)
|
24 |
+
self.srmodel.load_state_dict(loadnet['params_ema'], strict=True)
|
25 |
+
self.srmodel.eval()
|
26 |
+
self.srmodel = self.srmodel.to(self.device)
|
27 |
+
|
28 |
+
def tile_process(self, img):
|
29 |
+
"""It will first crop input images to tiles, and then process each tile.
|
30 |
+
Finally, all the processed tiles are merged into one images.
|
31 |
+
|
32 |
+
Modified from: https://github.com/ata4/esrgan-launcher
|
33 |
+
"""
|
34 |
+
batch, channel, height, width = img.shape
|
35 |
+
output_height = height * self.scale
|
36 |
+
output_width = width * self.scale
|
37 |
+
output_shape = (batch, channel, output_height, output_width)
|
38 |
+
|
39 |
+
# start with black image
|
40 |
+
output = img.new_zeros(output_shape)
|
41 |
+
tiles_x = math.ceil(width / self.tile_size)
|
42 |
+
tiles_y = math.ceil(height / self.tile_size)
|
43 |
+
|
44 |
+
# loop over all tiles
|
45 |
+
for y in range(tiles_y):
|
46 |
+
for x in range(tiles_x):
|
47 |
+
# extract tile from input image
|
48 |
+
ofs_x = x * self.tile_size
|
49 |
+
ofs_y = y * self.tile_size
|
50 |
+
# input tile area on total image
|
51 |
+
input_start_x = ofs_x
|
52 |
+
input_end_x = min(ofs_x + self.tile_size, width)
|
53 |
+
input_start_y = ofs_y
|
54 |
+
input_end_y = min(ofs_y + self.tile_size, height)
|
55 |
+
|
56 |
+
# input tile area on total image with padding
|
57 |
+
input_start_x_pad = max(input_start_x - self.tile_pad, 0)
|
58 |
+
input_end_x_pad = min(input_end_x + self.tile_pad, width)
|
59 |
+
input_start_y_pad = max(input_start_y - self.tile_pad, 0)
|
60 |
+
input_end_y_pad = min(input_end_y + self.tile_pad, height)
|
61 |
+
|
62 |
+
# input tile dimensions
|
63 |
+
input_tile_width = input_end_x - input_start_x
|
64 |
+
input_tile_height = input_end_y - input_start_y
|
65 |
+
tile_idx = y * tiles_x + x + 1
|
66 |
+
input_tile = img[:, :, input_start_y_pad:input_end_y_pad, input_start_x_pad:input_end_x_pad]
|
67 |
+
|
68 |
+
# upscale tile
|
69 |
+
try:
|
70 |
+
with torch.no_grad():
|
71 |
+
output_tile = self.srmodel(input_tile)
|
72 |
+
except RuntimeError as error:
|
73 |
+
print('Error', error)
|
74 |
+
return None
|
75 |
+
if tile_idx%10==0: print(f'\tTile {tile_idx}/{tiles_x * tiles_y}')
|
76 |
+
|
77 |
+
# output tile area on total image
|
78 |
+
output_start_x = input_start_x * self.scale
|
79 |
+
output_end_x = input_end_x * self.scale
|
80 |
+
output_start_y = input_start_y * self.scale
|
81 |
+
output_end_y = input_end_y * self.scale
|
82 |
+
|
83 |
+
# output tile area without padding
|
84 |
+
output_start_x_tile = (input_start_x - input_start_x_pad) * self.scale
|
85 |
+
output_end_x_tile = output_start_x_tile + input_tile_width * self.scale
|
86 |
+
output_start_y_tile = (input_start_y - input_start_y_pad) * self.scale
|
87 |
+
output_end_y_tile = output_start_y_tile + input_tile_height * self.scale
|
88 |
+
|
89 |
+
# put tile into output image
|
90 |
+
output[:, :, output_start_y:output_end_y,
|
91 |
+
output_start_x:output_end_x] = output_tile[:, :, output_start_y_tile:output_end_y_tile,
|
92 |
+
output_start_x_tile:output_end_x_tile]
|
93 |
+
return output
|
94 |
+
|
95 |
+
def process(self, img):
|
96 |
+
img = img.astype(np.float32) / 255.
|
97 |
+
img = torch.from_numpy(np.transpose(img[:, :, [2, 1, 0]], (2, 0, 1))).float()
|
98 |
+
img = img.unsqueeze(0).to(self.device)
|
99 |
+
|
100 |
+
if self.scale == 2:
|
101 |
+
mod_scale = 2
|
102 |
+
elif self.scale == 1:
|
103 |
+
mod_scale = 4
|
104 |
+
else:
|
105 |
+
mod_scale = None
|
106 |
+
if mod_scale is not None:
|
107 |
+
h_pad, w_pad = 0, 0
|
108 |
+
_, _, h, w = img.size()
|
109 |
+
if (h % mod_scale != 0):
|
110 |
+
h_pad = (mod_scale - h % mod_scale)
|
111 |
+
if (w % mod_scale != 0):
|
112 |
+
w_pad = (mod_scale - w % mod_scale)
|
113 |
+
img = F.pad(img, (0, w_pad, 0, h_pad), 'reflect')
|
114 |
+
|
115 |
+
try:
|
116 |
+
with torch.no_grad():
|
117 |
+
if self.tile_size > 0:
|
118 |
+
output = self.tile_process(img)
|
119 |
+
else:
|
120 |
+
output = self.srmodel(img)
|
121 |
+
del img
|
122 |
+
# remove extra pad
|
123 |
+
if mod_scale is not None:
|
124 |
+
_, _, h, w = output.size()
|
125 |
+
output = output[:, :, 0:h - h_pad, 0:w - w_pad]
|
126 |
+
output = output.data.squeeze().float().cpu().clamp_(0, 1).numpy()
|
127 |
+
output = np.transpose(output[[2, 1, 0], :, :], (1, 2, 0))
|
128 |
+
output = (output * 255.0).round().astype(np.uint8)
|
129 |
+
|
130 |
+
return output
|
131 |
+
except Exception as e:
|
132 |
+
print('sr failed:', e)
|
133 |
+
return None
|
sr_model/rrdbnet_arch.py
ADDED
@@ -0,0 +1,116 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import torch
|
2 |
+
from torch import nn as nn
|
3 |
+
from torch.nn import functional as F
|
4 |
+
|
5 |
+
from arch_util import default_init_weights, make_layer, pixel_unshuffle
|
6 |
+
|
7 |
+
|
8 |
+
class ResidualDenseBlock(nn.Module):
|
9 |
+
"""Residual Dense Block.
|
10 |
+
|
11 |
+
Used in RRDB block in ESRGAN.
|
12 |
+
|
13 |
+
Args:
|
14 |
+
num_feat (int): Channel number of intermediate features.
|
15 |
+
num_grow_ch (int): Channels for each growth.
|
16 |
+
"""
|
17 |
+
|
18 |
+
def __init__(self, num_feat=64, num_grow_ch=32):
|
19 |
+
super(ResidualDenseBlock, self).__init__()
|
20 |
+
self.conv1 = nn.Conv2d(num_feat, num_grow_ch, 3, 1, 1)
|
21 |
+
self.conv2 = nn.Conv2d(num_feat + num_grow_ch, num_grow_ch, 3, 1, 1)
|
22 |
+
self.conv3 = nn.Conv2d(num_feat + 2 * num_grow_ch, num_grow_ch, 3, 1, 1)
|
23 |
+
self.conv4 = nn.Conv2d(num_feat + 3 * num_grow_ch, num_grow_ch, 3, 1, 1)
|
24 |
+
self.conv5 = nn.Conv2d(num_feat + 4 * num_grow_ch, num_feat, 3, 1, 1)
|
25 |
+
|
26 |
+
self.lrelu = nn.LeakyReLU(negative_slope=0.2, inplace=True)
|
27 |
+
|
28 |
+
# initialization
|
29 |
+
default_init_weights([self.conv1, self.conv2, self.conv3, self.conv4, self.conv5], 0.1)
|
30 |
+
|
31 |
+
def forward(self, x):
|
32 |
+
x1 = self.lrelu(self.conv1(x))
|
33 |
+
x2 = self.lrelu(self.conv2(torch.cat((x, x1), 1)))
|
34 |
+
x3 = self.lrelu(self.conv3(torch.cat((x, x1, x2), 1)))
|
35 |
+
x4 = self.lrelu(self.conv4(torch.cat((x, x1, x2, x3), 1)))
|
36 |
+
x5 = self.conv5(torch.cat((x, x1, x2, x3, x4), 1))
|
37 |
+
# Emperically, we use 0.2 to scale the residual for better performance
|
38 |
+
return x5 * 0.2 + x
|
39 |
+
|
40 |
+
|
41 |
+
class RRDB(nn.Module):
|
42 |
+
"""Residual in Residual Dense Block.
|
43 |
+
|
44 |
+
Used in RRDB-Net in ESRGAN.
|
45 |
+
|
46 |
+
Args:
|
47 |
+
num_feat (int): Channel number of intermediate features.
|
48 |
+
num_grow_ch (int): Channels for each growth.
|
49 |
+
"""
|
50 |
+
|
51 |
+
def __init__(self, num_feat, num_grow_ch=32):
|
52 |
+
super(RRDB, self).__init__()
|
53 |
+
self.rdb1 = ResidualDenseBlock(num_feat, num_grow_ch)
|
54 |
+
self.rdb2 = ResidualDenseBlock(num_feat, num_grow_ch)
|
55 |
+
self.rdb3 = ResidualDenseBlock(num_feat, num_grow_ch)
|
56 |
+
|
57 |
+
def forward(self, x):
|
58 |
+
out = self.rdb1(x)
|
59 |
+
out = self.rdb2(out)
|
60 |
+
out = self.rdb3(out)
|
61 |
+
# Emperically, we use 0.2 to scale the residual for better performance
|
62 |
+
return out * 0.2 + x
|
63 |
+
|
64 |
+
class RRDBNet(nn.Module):
|
65 |
+
"""Networks consisting of Residual in Residual Dense Block, which is used
|
66 |
+
in ESRGAN.
|
67 |
+
|
68 |
+
ESRGAN: Enhanced Super-Resolution Generative Adversarial Networks.
|
69 |
+
|
70 |
+
We extend ESRGAN for scale x2 and scale x1.
|
71 |
+
Note: This is one option for scale 1, scale 2 in RRDBNet.
|
72 |
+
We first employ the pixel-unshuffle (an inverse operation of pixelshuffle to reduce the spatial size
|
73 |
+
and enlarge the channel size before feeding inputs into the main ESRGAN architecture.
|
74 |
+
|
75 |
+
Args:
|
76 |
+
num_in_ch (int): Channel number of inputs.
|
77 |
+
num_out_ch (int): Channel number of outputs.
|
78 |
+
num_feat (int): Channel number of intermediate features.
|
79 |
+
Default: 64
|
80 |
+
num_block (int): Block number in the trunk network. Defaults: 23
|
81 |
+
num_grow_ch (int): Channels for each growth. Default: 32.
|
82 |
+
"""
|
83 |
+
|
84 |
+
def __init__(self, num_in_ch, num_out_ch, scale=4, num_feat=64, num_block=23, num_grow_ch=32):
|
85 |
+
super(RRDBNet, self).__init__()
|
86 |
+
self.scale = scale
|
87 |
+
if scale == 2:
|
88 |
+
num_in_ch = num_in_ch * 4
|
89 |
+
elif scale == 1:
|
90 |
+
num_in_ch = num_in_ch * 16
|
91 |
+
self.conv_first = nn.Conv2d(num_in_ch, num_feat, 3, 1, 1)
|
92 |
+
self.body = make_layer(RRDB, num_block, num_feat=num_feat, num_grow_ch=num_grow_ch)
|
93 |
+
self.conv_body = nn.Conv2d(num_feat, num_feat, 3, 1, 1)
|
94 |
+
# upsample
|
95 |
+
self.conv_up1 = nn.Conv2d(num_feat, num_feat, 3, 1, 1)
|
96 |
+
self.conv_up2 = nn.Conv2d(num_feat, num_feat, 3, 1, 1)
|
97 |
+
self.conv_hr = nn.Conv2d(num_feat, num_feat, 3, 1, 1)
|
98 |
+
self.conv_last = nn.Conv2d(num_feat, num_out_ch, 3, 1, 1)
|
99 |
+
|
100 |
+
self.lrelu = nn.LeakyReLU(negative_slope=0.2, inplace=True)
|
101 |
+
|
102 |
+
def forward(self, x):
|
103 |
+
if self.scale == 2:
|
104 |
+
feat = pixel_unshuffle(x, scale=2)
|
105 |
+
elif self.scale == 1:
|
106 |
+
feat = pixel_unshuffle(x, scale=4)
|
107 |
+
else:
|
108 |
+
feat = x
|
109 |
+
feat = self.conv_first(feat)
|
110 |
+
body_feat = self.conv_body(self.body(feat))
|
111 |
+
feat = feat + body_feat
|
112 |
+
# upsample
|
113 |
+
feat = self.lrelu(self.conv_up1(F.interpolate(feat, scale_factor=2, mode='nearest')))
|
114 |
+
feat = self.lrelu(self.conv_up2(F.interpolate(feat, scale_factor=2, mode='nearest')))
|
115 |
+
out = self.conv_last(self.lrelu(self.conv_hr(feat)))
|
116 |
+
return out
|