-
Notifications
You must be signed in to change notification settings - Fork 2
/
Copy pathtest.py
204 lines (170 loc) · 10.9 KB
/
test.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
import os
import torch
import cv2 as cv
import numpy as np
from skimage.metrics import structural_similarity as SSIM
from skimage.metrics import peak_signal_noise_ratio as PSNR
import config
import models
def save_log(recon_root, name_dataset, name_image, psnr, ssim, manner, consecutive=True):
if not os.path.isfile(f"{recon_root}/Res_{name_dataset}_{manner}.txt"):
log = open(f"{recon_root}/Res_{name_dataset}_{manner}.txt", 'w')
log.write("=" * 120 + "\n")
log.close()
log = open(f"{recon_root}/Res_{name_dataset}_{manner}.txt", 'r+')
if consecutive:
old = log.read()
log.seek(0)
log.write(old)
log.write(
f"Res {name_image}: PSNR, {round(psnr, 2)}, SSIM, {round(ssim, 4)}\n")
log.close()
def testing(network, val, manner=config.para.manner, save_img=config.para.save):
"""
The pre-processing before TCS-Net's forward propagation and the testing platform.
"""
recon_root = "reconstructed_images"
if not os.path.isdir(recon_root):
os.mkdir(recon_root)
datasets = ["Set11"] if val else ["McM18", "LIVE29", "General100", "OST300"] # Names of folders (testing datasets)
with torch.no_grad():
for one_dataset in datasets:
if not os.path.isdir(f"{recon_root}/{one_dataset}"):
os.mkdir(f"{recon_root}/{one_dataset}")
test_dataset_path = f"dataset/test/{one_dataset}"
# Grey manner
if manner == "grey":
recon_dataset_path_grey = f"{recon_root}/{one_dataset}/grey/"
recon_dataset_path_grey_rate = f"{recon_root}/{one_dataset}/grey/{config.para.rate}"
if not os.path.isdir(recon_dataset_path_grey):
os.mkdir(recon_dataset_path_grey)
if not os.path.isdir(recon_dataset_path_grey_rate):
os.mkdir(recon_dataset_path_grey_rate)
sum_psnr, sum_ssim = 0., 0.
for _, _, images in os.walk(f"{test_dataset_path}/rgb/"):
for one_image in images:
name_image = one_image.split('.')[0]
x = cv.imread(f"{test_dataset_path}/rgb/{one_image}", flags=cv.IMREAD_GRAYSCALE)
x_ori = x
x = torch.from_numpy(x / 255.).float()
h, w = x.size()
lack = config.para.block_size - h % config.para.block_size if h % config.para.block_size != 0 else 0
padding_h = torch.zeros(lack, w)
expand_h = h + lack
inputs = torch.cat((x, padding_h), 0)
lack = config.para.block_size - w % config.para.block_size if w % config.para.block_size != 0 else 0
expand_w = w + lack
padding_w = torch.zeros(expand_h, lack)
inputs = torch.cat((inputs, padding_w), 1).unsqueeze(0).unsqueeze(0)
inputs = torch.cat(torch.split(inputs, split_size_or_sections=config.para.block_size, dim=3), dim=0)
inputs = torch.cat(torch.split(inputs, split_size_or_sections=config.para.block_size, dim=2), dim=0)
reconstruction, _ = network(inputs)
idx = expand_w // config.para.block_size
reconstruction = torch.cat(torch.split(reconstruction, split_size_or_sections=1 * idx, dim=0), dim=2)
reconstruction = torch.cat(torch.split(reconstruction, split_size_or_sections=1, dim=0), dim=3)
reconstruction = reconstruction.squeeze()[:h, :w]
x_hat = reconstruction.cpu().numpy() * 255.
x_hat = np.rint(np.clip(x_hat, 0, 255))
psnr = PSNR(x_ori, x_hat, data_range=255)
ssim = SSIM(x_ori, x_hat, data_range=255, multichannel=False)
sum_psnr += psnr
sum_ssim += ssim
if save_img:
cv.imwrite(f"{recon_dataset_path_grey_rate}/{name_image}.png", x_hat)
save_log(recon_root, one_dataset, name_image, psnr, ssim, f"_{config.para.rate}_{manner}")
save_log(recon_root, one_dataset, None,
sum_psnr / len(images), sum_ssim / len(images), f"_{config.para.rate}_{manner}_AVG", False)
print(
f"AVG RES: PSNR, {round(sum_psnr / len(images), 2)}, SSIM, {round(sum_ssim / len(images), 4)}")
if val:
return round(sum_psnr / len(images), 2), round(sum_ssim / len(images), 4)
# RGB manner
elif manner == "rgb":
recon_dataset_path_rgb = f"{recon_root}/{one_dataset}/rgb/"
recon_dataset_path_rgb_rate = f"{recon_root}/{one_dataset}/rgb/{config.para.rate}"
if not os.path.isdir(recon_dataset_path_rgb):
os.mkdir(recon_dataset_path_rgb)
if not os.path.isdir(recon_dataset_path_rgb_rate):
os.mkdir(recon_dataset_path_rgb_rate)
sum_psnr, sum_ssim = 0., 0.
for _, _, images in os.walk(f"{test_dataset_path}/rgb/"):
for one_image in images:
name_image = one_image.split('.')[0]
x = cv.imread(f"{test_dataset_path}/rgb/{one_image}")
x_ori = x
r, g, b = cv.split(x)
r = torch.from_numpy(np.asarray(r)).squeeze().float() / 255.
g = torch.from_numpy(np.asarray(g)).squeeze().float() / 255.
b = torch.from_numpy(np.asarray(b)).squeeze().float() / 255.
x = torch.from_numpy(x).float()
h, w = x.size()[0], x.size()[1]
lack = config.para.block_size - h % config.para.block_size if h % config.para.block_size != 0 else 0
padding_h = torch.zeros(lack, w)
expand_h = h + lack
inputs_r = torch.cat((r, padding_h), 0)
inputs_g = torch.cat((g, padding_h), 0)
inputs_b = torch.cat((b, padding_h), 0)
lack = config.para.block_size - w % config.para.block_size if w % config.para.block_size != 0 else 0
expand_w = w + lack
padding_w = torch.zeros(expand_h, lack)
inputs_r = torch.cat((inputs_r, padding_w), 1).unsqueeze(0).unsqueeze(0)
inputs_g = torch.cat((inputs_g, padding_w), 1).unsqueeze(0).unsqueeze(0)
inputs_b = torch.cat((inputs_b, padding_w), 1).unsqueeze(0).unsqueeze(0)
inputs_r = torch.cat(torch.split(inputs_r, split_size_or_sections=config.para.block_size, dim=3),
dim=0)
inputs_r = torch.cat(torch.split(inputs_r, split_size_or_sections=config.para.block_size, dim=2),
dim=0)
inputs_g = torch.cat(torch.split(inputs_g, split_size_or_sections=config.para.block_size, dim=3),
dim=0)
inputs_g = torch.cat(torch.split(inputs_g, split_size_or_sections=config.para.block_size, dim=2),
dim=0)
inputs_b = torch.cat(torch.split(inputs_b, split_size_or_sections=config.para.block_size, dim=3),
dim=0)
inputs_b = torch.cat(torch.split(inputs_b, split_size_or_sections=config.para.block_size, dim=2),
dim=0)
r_hat, _ = network(inputs_r.to(config.para.device))
g_hat, _ = network(inputs_g.to(config.para.device))
b_hat, _ = network(inputs_b.to(config.para.device))
idx = expand_w // config.para.block_size
r_hat = torch.cat(torch.split(r_hat, split_size_or_sections=1 * idx, dim=0), dim=2)
r_hat = torch.cat(torch.split(r_hat, split_size_or_sections=1, dim=0), dim=3)
r_hat = r_hat.squeeze()[:h, :w].cpu().numpy() * 255.
g_hat = torch.cat(torch.split(g_hat, split_size_or_sections=1 * idx, dim=0), dim=2)
g_hat = torch.cat(torch.split(g_hat, split_size_or_sections=1, dim=0), dim=3)
g_hat = g_hat.squeeze()[:h, :w].cpu().numpy() * 255.
b_hat = torch.cat(torch.split(b_hat, split_size_or_sections=1 * idx, dim=0), dim=2)
b_hat = torch.cat(torch.split(b_hat, split_size_or_sections=1, dim=0), dim=3)
b_hat = b_hat.squeeze()[:h, :w].cpu().numpy() * 255.
r_hat, g_hat, b_hat = np.rint(np.clip(r_hat, 0, 255)), \
np.rint(np.clip(g_hat, 0, 255)), \
np.rint(np.clip(b_hat, 0, 255))
reconstruction = cv.merge([r_hat, g_hat, b_hat])
psnr = PSNR(x_ori, reconstruction, data_range=255)
ssim = SSIM(x_ori, reconstruction, data_range=255, multichannel=True)
sum_psnr += psnr
sum_ssim += ssim
if save_img:
cv.imwrite(f"{recon_dataset_path_rgb_rate}/{name_image}.png",
(reconstruction))
save_log(recon_root, one_dataset, name_image, psnr, ssim, f"_{config.para.rate}_{manner}")
save_log(recon_root, one_dataset, None,
sum_psnr / len(images), sum_ssim / len(images), f"_{config.para.rate}_{manner}_AVG", False)
print(
f"AVG RES: PSNR, {round(sum_psnr / len(images), 2)}, SSIM, {round(sum_ssim / len(images), 4)}")
if val:
return round(sum_psnr / len(images), 2), round(sum_ssim / len(images), 4)
else:
raise NotImplemented(f"Error manner: {manner}.")
if __name__ == "__main__":
my_state_dict = config.para.my_state_dict
device = config.para.device
net = models.TCS_Net().eval().to(device)
if os.path.exists(my_state_dict):
if torch.cuda.is_available():
trained_model = torch.load(my_state_dict, map_location=device)
else:
raise Exception(f"No GPU.")
net.load_state_dict(trained_model)
else:
raise FileNotFoundError(f"Missing trained model of rate {config.para.rate}.")
testing(net, val=False, manner=config.para.manner, save_img=config.para.save)