forked from MrGemy95/visual-interaction-networks-pytorch
-
Notifications
You must be signed in to change notification settings - Fork 0
/
vin_dataset.py
161 lines (128 loc) · 7.36 KB
/
vin_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
from torch.utils.data import Dataset
import os
import numpy as np
import matplotlib.image as mpimg
import torch
class VinDataset(Dataset):
"""Face Landmarks dataset."""
def __init__(self, config, transform=None):
self.config = config
self.transform = transform
total_img = np.zeros((self.config.set_num, int(self.config.frame_num), self.config.height, self.config.weight,
self.config.col_dim))
for i in range(self.config.set_num):
for j in range(int(self.config.frame_num)):
total_img[i, j] = mpimg.imread(self.config.img_folder + "train/" + str(i) + '_' + str(j) + '.png')[:, :,
:self.config.col_dim]
total_data = np.zeros((self.config.set_num, int(self.config.frame_num), self.config.No * 5))
for i in range(self.config.set_num):
f = open(self.config.data_folder + "train/" + str(i) + ".csv", "r")
total_data[i] = [line[:-1].split(",") for line in f.readlines()]
total_data = np.reshape(total_data, [self.config.set_num, int(self.config.frame_num), self.config.No, 5])
# reshape img and data
input_img = np.zeros((self.config.set_num * (int(self.config.frame_num) - 14 + 1), 6, self.config.height,
self.config.weight, self.config.col_dim)
)
output_label = np.zeros((self.config.set_num * (int(self.config.frame_num) - 14 + 1), 8, self.config.No, 4)
)
output_S_label = np.zeros((self.config.set_num * (int(self.config.frame_num) - 14 + 1), 4, self.config.No, 4)
)
for i in range(self.config.set_num):
for j in range(int(self.config.frame_num) - 14 + 1):
input_img[i * (int(self.config.frame_num) - 14 + 1) + j] = total_img[i, j:j + 6]
output_label[i * (int(self.config.frame_num) - 14 + 1) + j] = np.reshape(total_data[i, j + 6:j + 14],
[8, self.config.No, 5])[
:, :, 1:5]
output_S_label[i * (int(self.config.frame_num) - 14 + 1) + j] = np.reshape(total_data[i, j + 2:j + 6],
[4, self.config.No, 5])[:, :,
1:5]
# shuffle
tr_data_num = int(len(input_img) * 1)
total_idx = np.arange(len(input_img))
np.random.shuffle(total_idx)
self.tr_data = input_img[total_idx]
self.tr_label = output_label[total_idx]
self.tr_S_label = output_S_label[total_idx]
def __len__(self):
return len(self.tr_data)
def __getitem__(self, idx):
sample = {'image': self.tr_data[idx], 'output_label': self.tr_label[idx],
'output_S_label': self.tr_S_label[idx]}
if self.transform:
sample = self.transform(sample),
return sample
class ToTensor(object):
"""Convert ndarrays in sample to Tensors."""
def __call__(self, sample):
image, output_label, output_S_label = sample['image'], sample['output_label'], sample[
'output_S_label']
# swap color axis because
# numpy image: H x W x C
# torch image: C X H X W
image = image.transpose((0, 3, 1, 2))
sample['image']=torch.from_numpy(image)
sample['output_label']=torch.from_numpy(output_label)
sample['output_S_label']=torch.from_numpy(output_S_label)
return sample
class ToTensorV2(object):
"""Convert ndarrays in sample to Tensors."""
def __call__(self, sample):
image, output_label, output_S_label = sample['image'], sample['output_label'], sample[
'output_S_label']
# swap color axis because
# numpy image: H x W x C
# torch image: C X H X W
image = image.transpose((0, 1, 4, 2,3))
sample['image']=torch.from_numpy(image)
sample['output_label']=torch.from_numpy(output_label)
sample['output_S_label']=torch.from_numpy(output_S_label)
return sample
class VinTestDataset(Dataset):
"""Face Landmarks dataset."""
def __init__(self, config, transform=None):
self.config = config
self.transform = transform
def __len__(self):
return 1
def __getitem__(self, idx):
total_img = np.zeros((self.config.set_num, int(self.config.frame_num), self.config.height, self.config.weight,
self.config.col_dim))
for i in range(self.config.set_num):
for j in range(int(self.config.frame_num)):
total_img[i, j] = mpimg.imread(self.config.img_folder + "train/" + str(i) + '_' + str(j) + '.png')[:, :,
:self.config.col_dim]
ts_img = np.zeros(
(1, int(self.config.frame_num), self.config.height, self.config.weight, self.config.col_dim),
dtype=float)
for i in range(1):
for j in range(int(self.config.frame_num)):
ts_img[i, j] = mpimg.imread(self.config.img_folder + "test/" + str(i) + "_" + str(j) + '.png')[:, :,
:self.config.col_dim]
ts_data = np.zeros((1, int(self.config.frame_num), self.config.No * 5), dtype=float)
for i in range(1):
f = open(self.config.data_folder + "test/" + str(i) + ".csv", "r")
ts_data[i] = [line[:-1].split(",") for line in f.readlines()]
# reshape img and data
input_img = np.zeros(
(1 * (int(self.config.frame_num) - 14 + 1), 6, self.config.height, self.config.weight,
self.config.col_dim),
dtype=float);
output_label = np.zeros((1 * (int(self.config.frame_num) - 14 + 1), 8, self.config.No, 4), dtype=float)
output_S_label = np.zeros((1 * (int(self.config.frame_num) - 14 + 1), 4, self.config.No, 4), dtype=float)
for i in range(1):
for j in range(int(self.config.frame_num) - 14 + 1):
input_img[i * (int(self.config.frame_num) - 14 + 1) + j] = total_img[i, j:j + 6]
output_label[i * (int(self.config.frame_num) - 14 + 1) + j] = np.reshape(ts_data[i, j + 6:j + 14],
[8, self.config.No, 5])[:,
:, 1:5]
output_S_label[i * (int(self.config.frame_num) - 14 + 1) + j] = np.reshape(ts_data[i, j + 2:j + 6],
[4, self.config.No, 5])[
:,
:, 1:5]
xy_origin = output_label[:(int(self.config.frame_num) - 14 + 1 - 4 + 1), 0, :, 0:2]
xy_estimated = np.zeros((self.config.roll_num, self.config.No, 2), dtype=float)
sample = {'image': input_img, 'output_label': output_label,
'output_S_label': output_S_label,'xy_origin':xy_origin,'xy_estimated':xy_estimated}
if self.transform:
sample = self.transform(sample),
return sample