-
Notifications
You must be signed in to change notification settings - Fork 10
/
Copy pathdataloader.py
executable file
·166 lines (133 loc) · 6.19 KB
/
dataloader.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
#%%
from time import time
import numpy as np
import random
import utils.dataloader_util as dataloader_util
import os, cv2
from torch.utils.data.dataset import Dataset
from glob import glob
from torch.utils.data import DataLoader
seed = 2020
np.random.seed(seed)
random.seed(seed)
# TODO reorganize the argument
class SfPDataset(Dataset):
def __init__(self,
dataroot,
train = 1,
split = "inter",
use_deepsfp = True,
use_sfpwild = True,
interpolated_normal = False,
fast = False,
with_viewing_dir = False,
crop_interval = 32
) :
super().__init__()
self.dataroot = dataroot
self.split = split
self.use_deepsfp = use_deepsfp
self.use_sfpwild = use_sfpwild
self.interpolated_normal = interpolated_normal
self.fast = fast
self.with_viewing_dir = with_viewing_dir
self.train = train
self.crop_interval = crop_interval
if self.split == "inter":
train_pol_paths, train_normal_paths, train_mask_paths, \
test_pol_paths, test_normal_paths, test_mask_paths = dataloader_util.inter_scenes_split(self)
if train:
self.pol_paths = train_pol_paths
self.normal_paths = train_normal_paths
self.mask_paths = train_mask_paths
else:
self.pol_paths = test_pol_paths
self.normal_paths = test_normal_paths
self.mask_paths = test_mask_paths
print("training:", train)
print("mask_paths", self.mask_paths[0])
print("normal_paths", self.normal_paths[0])
print("pol_paths", self.pol_paths[0])
self.image_coordinate = dataloader_util.get_coordinate(np.load(train_pol_paths[0]), viewing_dir=self.with_viewing_dir)
self.viewing_direction = dataloader_util.get_vd()
def __getitem__(self, index):
'''
cpu only do easy job like
np.load input file, crop the file, random flip
'''
# import pdb; pdb.set_trace()
net_in, vis_camera_normal, net_gt, net_mask, net_coordinate, viewing_direction \
= dataloader_util.load_paired_data( \
self.pol_paths[index], self.normal_paths[index], self.mask_paths[index], self.image_coordinate, self.viewing_direction, crop_interval=self.crop_interval)
if self.train:
net_in, vis_camera_normal, net_gt, net_mask, net_coordinate, viewing_direction \
= dataloader_util.crop_augmentation_list( \
[net_in, vis_camera_normal, net_gt, net_mask, net_coordinate, viewing_direction])
sample_list = net_mask, net_in, vis_camera_normal, net_gt, net_coordinate, viewing_direction
tensor_list = dataloader_util.ToTensor(sample_list)
return tensor_list
def __len__(self):
return len(self.pol_paths)
class SfPTestDataset(Dataset):
def __init__(self,
dataroot,
txt_path=None,
with_viewing_dir=False,
use_deepsfp=None,
crop_interval=32
) -> None:
super().__init__()
self.dataroot = dataroot
self.with_viewing_dir = with_viewing_dir
if txt_path:
# txt_path = txt_path.replace("/home/chenyang/disk1/iccv2021-sfp-wild/data/iccv2021", dataroot)
print(txt_path)
with open(txt_path, 'r') as f:
self.pol_paths = [line[:-1].replace("DATAROOT", dataroot) for line in f.readlines()]
else:
if use_deepsfp:
self.pol_paths = sorted(glob("{}/eccv2020/test/*/*polar.npy".format(dataroot)))[1:]
print("{}/eccv2020/test/*/*polar.npy".format(dataroot))
else:
self.pol_paths = sorted(glob("{}/*png".format(dataroot)))
self.image_coordinate = dataloader_util.get_coordinate(np.load(self.pol_paths[0]), viewing_dir=self.with_viewing_dir)
self.viewing_direction = dataloader_util.get_vd()
self.crop_interval=crop_interval
def __getitem__(self, index):
'''
cpu only do easy job like
np.load input file, crop the file, random flip
'''
net_in = np.load(self.pol_paths[index])
h, w = net_in.shape[:2]
h = h // self.crop_interval * self.crop_interval
w = w // self.crop_interval * self.crop_interval
net_in = net_in[:h,:w]
net_coordinate = self.image_coordinate[0,:h,:w]
viewing_direction = self.viewing_direction[0,:h,:w]
sample_list = net_in, net_coordinate, viewing_direction
tensor_list = dataloader_util.ToTensor(sample_list)
return tensor_list
def __len__(self):
return len(self.pol_paths)
import torch
def create_dataloader(args):
print('> Loading datasets ...')
sfp_train_dataset = SfPDataset(args.dataroot,
split=args.split,
use_deepsfp = args.use_deepsfp,
use_sfpwild = args.use_sfpwild,
interpolated_normal=args.interpolated_normal,
train=1, fast=(args.training_mode=="fast"),
crop_interval=args.crop_interval)
sfp_test_dataset = SfPDataset(args.dataroot,
split=args.split,
use_deepsfp = args.use_deepsfp,
use_sfpwild = args.use_sfpwild,
interpolated_normal=args.interpolated_normal,
train=0, fast=(args.training_mode=="fast"),
crop_interval=args.crop_interval)
return sfp_train_dataset, sfp_test_dataset
if __name__=="__main__":
sfp_train_dataset = SfPDataset("./data/iccv2021", train=1, with_flip_aug=True)
return_list = sfp_train_dataset.__getitem__(0)