-
Notifications
You must be signed in to change notification settings - Fork 3
/
Copy pathdataset.py
211 lines (168 loc) · 7.3 KB
/
dataset.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
205
206
207
208
209
210
211
import os
import torch.utils.data as data
import cv2
import utils
import numpy as np
import random
from matlab_imresize import imresize
from scipy import ndimage
def make_dataset(img_path, Q_list, npy, sr_factor):
img_paths_list = []
target_path = os.path.join(img_path, 'origin_npy' if npy else 'origin')
for root, _, names in os.walk(target_path):
for name in names:
target_name = os.path.join(target_path, name)
for Q in Q_list:
jpeg_ = Q+'_npy' if npy else Q
input_path = img_path + '/x%d/'%(sr_factor) + jpeg_
name, _ = name.split('.')
save_name = name + '.png'
if Q == 'jpeg0':
name = name + '.npy' if npy else name + '.png' #When Q = 0, it means images are just downsampled by bicubic.
else:
if npy:
name = name + '.npy'
elif 'jpeg' in Q:
name = name + '.jpg'
elif 'webp' in Q:
name = name + '.webp'
input_name = os.path.join(input_path, name)
img_paths_list.append({'input': input_name, 'target': target_name, 'save': save_name})
return img_paths_list
class DatasetTrain(data.Dataset):
def __init__(self, opt, target_down=False):
super(DatasetTrain, self).__init__()
self.sr_factor = opt.sr_factor
self.img_path = opt.train_root
self.Q_list = opt.Q_list
self.npy_reader = opt.npy_reader
self.train_path = make_dataset(self.img_path, self.Q_list, self.npy_reader, self.sr_factor)
if opt.n_colors == 1:
self.rgb = False
elif opt.n_colors == 3:
self.rgb = True
else:
raise ValueError('n_colors must be 1 or 3.')
self.patch_size = opt.patch_size
self.rgb_range = opt.rgb_range
self.target_down = target_down
def __getitem__(self, index):
path = self.train_path[index]
images = {}
if self.npy_reader:
input_ = np.load(path['input'], allow_pickle=False)
target_ = np.load(path['target'], allow_pickle=False)
target_ = utils.modcrop(target_, self.sr_factor)
else:
input_ = cv2.imread(path['input'])
input_ = cv2.cvtColor(input_, cv2.COLOR_BGR2RGB)
target_ = cv2.imread(path['target'])
target_ = utils.modcrop(target_, self.sr_factor)
target_ = cv2.cvtColor(target_, cv2.COLOR_BGR2RGB)
# for i in range(10):
# subim_in, subim_tar = get_patch(input_, target_, self.patch_size, self.sr_factor)
# win_mean = ndimage.uniform_filter(subim_in[:, :, 0], (5, 5))
# win_sqr_mean = ndimage.uniform_filter(subim_in[:, :, 0]**2, (5, 5))
# win_var = win_sqr_mean - win_mean**2
#
# if np.sum(win_var) / (win_var.shape[0]*win_var.shape[1]) > 30:
# break
subim_in, subim_tar = get_patch(input_, target_, self.patch_size, self.sr_factor)
if not self.rgb:
subim_in = utils.rgb2ycbcr(subim_in)
subim_tar = utils.rgb2ycbcr(subim_tar)
subim_in = np.expand_dims(subim_in[:, :, 0], 2)
subim_tar = np.expand_dims(subim_tar[:, :, 0], 2)
if self.target_down:
subim_target_down = imresize(subim_tar, scalar_scale=1 / self.sr_factor)
subim_target_down = utils.np2tensor(subim_target_down, self.rgb_range)
images.update({'target_down': subim_target_down})
subim_in = utils.np2tensor(subim_in, self.rgb_range)
subim_tar = utils.np2tensor(subim_tar, self.rgb_range)
images.update({'input': subim_in, 'target': subim_tar})
return images
def __len__(self):
return len(self.train_path)
class DatasetReal(data.Dataset):
def __init__(self, opt):
super(DatasetReal, self).__init__()
self.sr_factor = opt.sr_factor
self.img_path = opt.test_root
self.img_paths_list = []
for root, _, names in os.walk(self.img_path):
for name in names:
input_name = os.path.join(self.img_path, name)
name, _ = name.split('.')
save_name = name + '.png'
self.img_paths_list.append({'input': input_name, 'save': save_name})
self.rgb_range = opt.rgb_range
def __getitem__(self, index):
path = self.img_paths_list[index]
print(path)
images = {}
images.update({'name': path['save']})
input_ = cv2.imread(path['input'])
input_ = cv2.cvtColor(input_, cv2.COLOR_BGR2RGB)
input_ = utils.np2tensor(input_, self.rgb_range)
images.update({'input': input_, 'target': input_})
return images
def __len__(self):
return len(self.img_paths_list)
class DatasetTest(data.Dataset):
def __init__(self, opt, target_down=False, Q_list=None):
super(DatasetTest, self).__init__()
self.sr_factor = opt.sr_factor
self.img_path = opt.test_root
if Q_list is None:
self.Q_list = opt.Q_list
else:
self.Q_list = Q_list
self.npy_reader = False # when testing, there's no need to use npy data.
if opt.n_colors == 1:
self.rgb = False
elif opt.n_colors == 3:
self.rgb = True
else:
raise ValueError('n_colors must be 1 or 3.')
self.test_path = make_dataset(self.img_path, self.Q_list, self.npy_reader, self.sr_factor)
self.rgb_range = opt.rgb_range
self.target_down = target_down
def __getitem__(self, index):
path = self.test_path[index]
images = {}
images.update({'name': path['save']})
input_ = cv2.imread(path['input'])
input_ = cv2.cvtColor(input_, cv2.COLOR_BGR2RGB)
target_ = cv2.imread(path['target'])
target_ = utils.modcrop(target_, self.sr_factor)
target_ = cv2.cvtColor(target_, cv2.COLOR_BGR2RGB)
if not self.rgb:
input_out = np.copy(input_)
input_out = utils.np2tensor(input_out, self.rgb_range)
# print(input_out)
input_ = utils.rgb2ycbcr(input_)
input_cbcr = input_[:, :, 1:]
input_ = np.expand_dims(input_[:, :, 0], 2)
input_cbcr = utils.np2tensor(input_cbcr, self.rgb_range)
images.update({'input_cbcr': input_cbcr, 'input_rgb': input_out})
if self.target_down:
target_down = imresize(target_, scalar_scale=1/self.sr_factor)
target_down = utils.np2tensor(target_down, self.rgb_range)
images.update({'target_down': target_down})
input_ = utils.np2tensor(input_, self.rgb_range)
target_ = utils.np2tensor(target_, self.rgb_range)
images.update({'input': input_, 'target': target_})
return images
def __len__(self):
return len(self.test_path)
def get_patch(img_in, img_tar, patch_size, scale):
ih, iw = img_in.shape[:2]
# p = scale
ip = patch_size
tp = ip * scale
ix = random.randrange(0, iw - ip + 1)
iy = random.randrange(0, ih - ip + 1)
tx, ty = scale * ix, scale * iy
img_in = img_in[iy:iy + ip, ix:ix + ip, :]
img_tar = img_tar[ty:ty + tp, tx:tx + tp, :]
return img_in, img_tar