-
Notifications
You must be signed in to change notification settings - Fork 13
/
DatasetLoader.py
191 lines (159 loc) · 7.15 KB
/
DatasetLoader.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
import torch
from torchvision import datasets, transforms
import torch.utils.data as data
import os
from PIL import Image
import numpy as np
from torch.utils.data.sampler import SubsetRandomSampler, SequentialSampler
base_transform = transforms.Compose([
transforms.ToTensor(),
# transforms.Normalize([0.5] * 3, [0.5] * 3)
])
norm_transform = transforms.Compose([
transforms.Normalize([0.5] * 3, [0.5] * 3)
])
inverse_transform = transforms.Compose([transforms.Normalize(mean=[0., 0., 0.],
std=[1 / 0.5] * 3),
transforms.Normalize(mean=[-0.5] * 3,
std=[1., 1., 1.]),
])
upscale_factor = 4
def image_loader(path):
img = Image.open(path)
img_tensor = base_transform(img)
return img_tensor
def lr_image_loader(path, upscale_factor):
""" Downscale an HR image to an LR image. """
img = Image.open(path)
width, height = img.size
tw = width//upscale_factor
th = height//upscale_factor
#print((tw,th))
img = img.resize(size = (tw,th), resample = Image.BILINEAR)
#print("lr:",img.size)
img_tensor = base_transform(img)
#print(img_tensor.shape)
return img_tensor
class FRDataset(data.Dataset):
def __init__(self, hr_dir, upscale_factor):
self.file_hr_dir = hr_dir
self.transform = base_transform
self.upscale_factor = upscale_factor
# self.image_loader = image_loader()
self.hr_frames_set = os.listdir(self.file_hr_dir)
def __getitem__(self, index):
def get_from_set(dir, frame_set):
frames = frame_set[index] # 0266
# print(f'frame is {frames}, typ is {type(frames)}')
# frame_tensor = torch.Tensor(size=(frame_counter, 3, self.height, self.weight))
frame_tensor = []
# file_dir_frames = self.file_dir + frames
file_dir_frames = os.path.join(dir, frames)
imgs_path = os.listdir(file_dir_frames)
imgs_path.sort()
i = 0
for img in imgs_path:
final_path = file_dir_frames + "/" + img
# final_path = '/'.os.listdir(file_dir_frames,img)
img_tensor = image_loader(final_path)
# print(img_tensor.size())
frame_tensor.append(img_tensor)
i = i + 1
res = torch.stack(frame_tensor, dim=0)
# print(f'res has shape {res.shape}')
return res
def get_lr_from_set(dir, frame_set, upscale_factor):
frames = frame_set[index] # 0266
# print(f'frame is {frames}, typ is {type(frames)}')
# frame_tensor = torch.Tensor(size=(frame_counter, 3, self.height, self.weight))
frame_tensor = []
# file_dir_frames = self.file_dir + frames
file_dir_frames = os.path.join(dir, frames)
imgs_path = os.listdir(file_dir_frames)
imgs_path.sort()
i = 0
for img in imgs_path:
final_path = file_dir_frames + "/" + img
# final_path = '/'.os.listdir(file_dir_frames,img)
img_tensor = lr_image_loader(final_path,upscale_factor)
# print(img_tensor.size())
frame_tensor.append(img_tensor)
i = i + 1
res = torch.stack(frame_tensor, dim=0)
# print(f'res has shape {res.shape}')
return res
return get_lr_from_set(self.file_hr_dir, self.hr_frames_set,self.upscale_factor), \
get_from_set(self.file_hr_dir, self.hr_frames_set)
def __len__(self):
return len(self.hr_frames_set)
# # this returns the basic infomation of the dataset.
# def touch(self):
class loader_wrapper(object):
def __init__(self, loader):
self.loader = loader
def __iter__(self):
for lr_img, hr_img in self.loader:
yield lr_img.permute(1, 0, 2, 3, 4), hr_img.permute(1, 0, 2, 3, 4)
def __len__(self):
return len(self.loader)
def get_data_loaders(batch, shuffle_dataset=True, dataset_size=0, validation_split=0.2, fixedIndices=-1):
# batch = 4 # batch size of the data every time for training
# batch_number = 100000 # number of batches, so we totally have batch_number * batch images
# HR_height = height
# HR_width = width
#
# LR_height = HR_height // SRFactor
# LR_width = HR_width // SRFactor
train_dir_HR = 'Data/HR'
FRData = FRDataset(hr_dir=train_dir_HR, upscale_factor=upscale_factor)
# data_loader_LR = data.DataLoader(FRData_LR, batch_size = batch, shuffle = True)
# data_loader_HR = data.DataLoader(FRData_HR, batch_size = batch, shuffle = True)
# print(data_loader[0].size())
random_seed = 42
if dataset_size == 0:
dataset_size = len(FRData)
print("Dataset size:", len(FRData))
indices = list(range(dataset_size))
split = int(np.floor(validation_split * dataset_size))
if shuffle_dataset:
np.random.seed(random_seed)
np.random.shuffle(indices)
indices = [fixedIndices] if fixedIndices != -1 else indices
train_indices, val_indices = indices[split:], indices[:split]
print("Training/Validation split: %s/%s", (1-validation_split)*100, validation_split*100)
print("Training samples chosen:", train_indices)
print("Validation samples chosen:", val_indices)
train_sampler = SubsetRandomSampler(train_indices)
print("Training set size:", len(train_sampler))
valid_sampler = SubsetRandomSampler(val_indices)
print("Validation set size:", len(valid_sampler))
train_loader = torch.utils.data.DataLoader(FRData, batch_size=batch, sampler=train_sampler, drop_last=True)
validation_loader = torch.utils.data.DataLoader(FRData, batch_size=batch, sampler=valid_sampler, drop_last=True)
train_loader = loader_wrapper(train_loader)
validation_loader = loader_wrapper(validation_loader)
return train_loader, validation_loader
if __name__ == "__main__":
train, val = get_data_loaders(4)
for lr_img, hr_img in train:
print(f'lr_img shape is {lr_img.shape}, hr_img shape is {hr_img.shape}')
break
# class TestFRVSR(unittest.TestCase):
# def TestGetDataLoader(self):
#
# for i_batch, sample_batched in enumerate(zip(train_loader_LR, train_loader_HR)):
# #print(sample_batched)
# #print(data_loader_HR[i_batch].size())
# permuted_LR_data = sample_batched[0].permute(1, 0, 2, 3, 4)
# permuted_HR_data = sample_batched[1].permute(1, 0, 2, 3, 4) #labels
# #print(permuted_data.contiguous())
# print("LR:",permuted_LR_data.size())
# print("HR:",permuted_HR_data.size())
#
# for j_batch, sample_batched in enumerate(zip(validation_loader_LR, validation_loader_HR)):
# #print(sample_batched)
# #print(data_loader_HR[i_batch].size())
# permuted_LR_data = sample_batched[0].permute(1, 0, 2, 3, 4)
# permuted_HR_data = sample_batched[1].permute(1, 0, 2, 3, 4) #labels
# #print(permuted_data.contiguous())
# print("LR:",permuted_LR_data.size())
# print("HR:",permuted_HR_data.size())