-
Notifications
You must be signed in to change notification settings - Fork 17
/
data.py
188 lines (143 loc) · 6.18 KB
/
data.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
from __future__ import print_function, division
import os
import cv2
import torch
import random
from skimage import io, transform
from PIL import Image
import numpy as np
import matplotlib.pyplot as plt
from torch.utils.data import Dataset, DataLoader
from torchvision import transforms, utils
from utils import resize, crop, adjust_brightness, adjust_saturation
from utils import adjust_contrast, adjust_hue
def pil_loader(path):
# open path as file to avoid ResourceWarning
# (https://github.com/python-pillow/Pillow/issues/835)
with open(path, 'rb') as f:
with Image.open(f) as img:
return img.convert('RGB')
def default_loader(path):
return pil_loader(path)
class VisionTouchDataset(Dataset):
def __init__(self, phase, data_lst_file,
w_timewindow,
trans_des=None, trans_lowres=None,
trans_to_tensor=None,
scale_size=None, crop_size=None,
brightness=None, contrast=None, saturation=None, hue=None,
loader=default_loader):
self.phase = phase
self.recs = open(data_lst_file, 'r').readlines()
self.w_timewindow = w_timewindow
self.trans_des = trans_des
self.trans_lowres = trans_lowres
self.trans_to_tensor = trans_to_tensor
self.loader = loader
self.scale_size = scale_size
self.crop_size = crop_size
self.brightness = brightness
self.contrast = contrast
self.saturation = saturation
self.hue = hue
def __len__(self):
return len(self.recs)
def variance_of_laplacian(self, image):
return cv2.Laplacian(image, cv2.CV_64F).var() ** 2
def cvt_rgb2gray(self, image):
return cv2.cvtColor(np.array(image).copy(), cv2.COLOR_RGB2GRAY)
def calc_weight(self, ref_des, des):
gray_ref = np.array(self.cvt_rgb2gray(ref_des)).astype(np.float)
gray = np.array(self.cvt_rgb2gray(des)).astype(np.float)
return self.variance_of_laplacian(gray - gray_ref)
def get_crop_params(self, phase, img, crop_size):
w, h = img.size
th, tw = crop_size, crop_size # hack for now
if phase == 'train':
if w == tw and h == th:
return 0, 0, h, w
i = random.randint(0, h - th)
j = random.randint((w - tw) / 2. - 8, (w - tw) / 2. + 8)
else:
i = int(round((h - th) / 2.))
j = int(round((w - tw) / 2.))
return i, j, th, tw
def resize_and_crop(self, phase, srcs, scale_size, crop_size):
len_srcs = len(srcs)
# resize the images
for i in range(len_srcs):
srcs[i] = resize(srcs[i], scale_size)
crop_params = self.get_crop_params(phase, srcs[0], crop_size)
# crop the images
for i in range(len_srcs):
srcs[i] = crop(srcs[i], crop_params[0], crop_params[1], crop_params[2], crop_params[3])
return srcs
def colorjitter(self, srcs, brightness, contrast, saturation, hue):
len_srcs = len(srcs)
brightness_factor = np.random.uniform(max(0, 1 - brightness), 1 + brightness)
contrast_factor = np.random.uniform(max(0, 1 - contrast), 1 + contrast)
saturation_factor = np.random.uniform(max(0, 1 - saturation), 1 + saturation)
hue_factor = np.random.uniform(-hue, hue)
for i in range(len_srcs):
srcs[i] = adjust_brightness(srcs[i], brightness_factor)
srcs[i] = adjust_contrast(srcs[i], contrast_factor)
srcs[i] = adjust_saturation(srcs[i], saturation_factor)
srcs[i] = adjust_hue(srcs[i], hue_factor)
return srcs
def __getitem__(self, idx):
ref_src, ref_des, src, des, \
src_pre_0, src_pre_1, \
src_nxt_0, src_nxt_1 = self.recs[idx].strip().split(" ")
ref_src = self.loader(ref_src)
ref_des = self.loader(ref_des)
src = self.loader(src)
src_rgb = src.copy()
des = self.loader(des)
if self.w_timewindow:
src_pre_0 = self.loader(src_pre_0)
src_pre_1 = self.loader(src_pre_1)
src_nxt_0 = self.loader(src_nxt_0)
src_nxt_1 = self.loader(src_nxt_1)
srcs = [ref_src, src, src_pre_0, src_nxt_1, src_pre_1, src_nxt_0]
else:
srcs = [ref_src, src]
# transform src
srcs = self.resize_and_crop(self.phase, srcs, self.scale_size, self.crop_size)
if self.phase == 'train':
srcs = self.colorjitter(srcs, self.brightness, self.contrast,
self.saturation, self.hue)
srcs_lowres = []
for i in range(len(srcs)):
srcs_lowres += [self.trans_lowres(srcs[i])]
if self.w_timewindow:
for i in range(1, len(srcs)):
srcs[i] = self.cvt_rgb2gray(srcs[i])
srcs_lowres[i] = self.cvt_rgb2gray(srcs_lowres[i])
ref_src = srcs[0]
ref_src_lowres = srcs_lowres[0]
src = np.stack((srcs[1], srcs[2], srcs[3], srcs[4], srcs[5]), axis=-1)
src_lowres = np.stack((srcs_lowres[1], srcs_lowres[2],
srcs_lowres[3], srcs_lowres[4],
srcs_lowres[5]), axis=-1)
else:
ref_src = srcs[0]
ref_src_lowres = srcs_lowres[0]
src = srcs[1]
src_lowres = srcs_lowres[1]
# transform des
ref_des = self.trans_des(ref_des)
ref_des_lowres = self.trans_lowres(ref_des)
des = self.trans_des(des)
des_lowres = self.trans_lowres(des)
# transform all images to torch tensor
ref_src = self.trans_to_tensor(ref_src)
ref_src_lowres = self.trans_to_tensor(ref_src_lowres)
src = self.trans_to_tensor(src)
src_lowres = self.trans_to_tensor(src_lowres)
src_rgb = self.trans_to_tensor(src_rgb)
ref_des = self.trans_to_tensor(ref_des)
ref_des_lowres = self.trans_to_tensor(ref_des_lowres)
des = self.trans_to_tensor(des)
des_lowres = self.trans_to_tensor(des_lowres)
return ref_src_lowres, ref_des_lowres, src_lowres, des_lowres, \
ref_src, ref_des, src, src_rgb