/
data_loader.py
188 lines (159 loc) · 6.05 KB
/
data_loader.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
#%%
import os
import matplotlib.pyplot as plt
import numpy as np
import torch
import cv2
from torchvision import transforms
import random
from PIL import Image
from torchvision.transforms import Compose, ToTensor, Normalize
import torchvision.transforms.functional as F_vision
import torch.nn.functional as F
import pandas as pd
import scipy.io as scio
class CrowdDataset(torch.utils.data.Dataset):
'''
CrowdDataset
'''
def __init__(self, root, phase, main_transform=None, img_transform=None, dmap_transform=None):
'''
root: the root path of dataset.
phase: train or test.
main_transform: transforms on both image and density map.
img_transform: transforms on image.
dmap_transform: transforms on densitymap.
'''
self.img_path = os.path.join(root, phase+'_frame/')
self.dmap_path = os.path.join(root, phase+'_perspective/')
self.data_files = [filename for filename in os.listdir(self.img_path)
if os.path.isfile(os.path.join(self.img_path, filename))]
self.main_transform = main_transform
self.img_transform = img_transform
self.dmap_transform = dmap_transform
self.phase=phase
def __len__(self):
return len(self.data_files)
def __getitem__(self, index):
index = index % len(self.data_files)
fname = self.data_files[index]
img, dmap = self.read_image_and_dmap(fname)
if self.main_transform is not None:
img, dmap = self.main_transform((img, dmap))
if self.img_transform is not None:
img = self.img_transform(img)
if self.dmap_transform is not None:
dmap = self.dmap_transform(dmap)
# if self.phase=="test":
# pmap_pred=dmap.numpy()[0]
# # pmap_pred[pmap_pred>1]=1
# # pmap_pred[pmap_pred<0]=0
# pmap_pred=np.asarray(pmap_pred*255, dtype=np.uint8)
# feature_img = cv2.applyColorMap(pmap_pred, cv2.COLORMAP_JET)
# cv2.imwrite("GT/%d.png"%(index), feature_img)
return {'image': img, 'densitymap': dmap}
def read_image_and_dmap(self, fname):
img = Image.open(os.path.join(self.img_path, fname))
if img.mode == 'L':
# print('There is a grayscale image.')
img = img.convert('RGB')
image_id=fname[:6]
dmap = scio.loadmat(os.path.join(
self.dmap_path, image_id + '.mat'))
dmap = dmap['pMap'].astype(np.float32, copy=False)
dmap=dmap/255.0
dmap = Image.fromarray(dmap)
return img, dmap
def create_train_dataloader(root, use_flip, batch_size):
'''
Create train dataloader.
root: the dataset root.
use_flip: True or false.
batch size: the batch size.
'''
main_trans_list = []
if use_flip:
main_trans_list.append(RandomHorizontalFlip())
main_trans_list.append(PairedCrop())
main_trans = Compose(main_trans_list)
img_trans = Compose([ToTensor(), Normalize(mean=[0.5,0.5,0.5],std=[0.225,0.225,0.225])])
dmap_trans = ToTensor()
dataset = CrowdDataset(root=root, phase='train', main_transform=main_trans,
img_transform=img_trans,dmap_transform=dmap_trans)
dataloader = torch.utils.data.DataLoader(dataset,batch_size=batch_size,shuffle=True,num_workers=12)
return dataloader
def create_test_dataloader(root):
'''
Create train dataloader.
root: the dataset root.
'''
main_trans_list = []
main_trans_list.append(PairedCrop())
main_trans = Compose(main_trans_list)
img_trans = Compose([ToTensor(), Normalize(mean=[0.5,0.5,0.5],std=[0.225,0.225,0.225])])
dmap_trans = ToTensor()
dataset = CrowdDataset(root=root, phase='test', main_transform=main_trans,
img_transform=img_trans,dmap_transform=dmap_trans)
dataloader = torch.utils.data.DataLoader(dataset,batch_size=1,shuffle=False)
return dataloader
#----------------------------------#
# Transform code #
#----------------------------------#
class RandomHorizontalFlip(object):
'''
Random horizontal flip.
prob = 0.5
'''
def __call__(self, img_and_dmap):
'''
img: PIL.Image
dmap: PIL.Image
'''
img, dmap = img_and_dmap
if random.random() < 0.5:
return (img.transpose(Image.FLIP_LEFT_RIGHT), dmap.transpose(Image.FLIP_LEFT_RIGHT))
else:
return (img, dmap)
class PairedCrop(object):
'''
Paired Crop for both image and its density map.
Note that due to the maxpooling in the nerual network,
we must promise that the size of input image is the corresponding factor.
'''
def __init__(self, factor=16):
self.factor = factor
@staticmethod
def get_params(img, factor):
w, h = img.size
if w % factor == 0 and h % factor == 0:
return 0, 0, h, w
else:
return 0, 0, h - (h % factor), w - (w % factor)
def __call__(self, img_and_dmap):
'''
img: PIL.Image
dmap: PIL.Image
'''
img, dmap = img_and_dmap
i, j, th, tw = self.get_params(img, self.factor)
img = F_vision.crop(img, i, j, th, tw)
dmap = F_vision.crop(dmap, i, j, th, tw)
img =img.resize((256, 256),Image.ANTIALIAS)
dmap =dmap.resize((256, 256),Image.ANTIALIAS)
return (img, dmap)
def find_nearest(array, value):
array = np.asarray(array)
idx = (np.abs(array - value)).argmin()
return array[idx],idx
# testing code
#[6.60696751e-05 1.53686842e-04 3.03317757e-04 5.70904907e-04 1.09181233e-03 2.36344733e-03 4.39890302e-03 7.51745701e-03]
#[54 ,93,75,25,56,22]
if __name__ == "__main__":
root = '/home/wangjian/CrowdCounting/WorldExpo/'
#dataloader = create_train_dataloader(root,False,1)
dataloader = create_test_dataloader(root)
vector_np=[]
for i, data in enumerate(dataloader):
image = data['image']
densitymap = data['densitymap']
#print(image.shape,densitymap.shape)