-
Notifications
You must be signed in to change notification settings - Fork 151
/
datasets.py
273 lines (213 loc) · 8.87 KB
/
datasets.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
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
import collections
import glob
import os
import os.path as osp
import numpy as np
import torch
from PIL import Image
from PIL import ImageOps
from torch.utils import data
from transform import HorizontalFlip, VerticalFlip
class ConcatDataset(torch.utils.data.Dataset):
def __init__(self, *datasets):
self.datasets = datasets
def __getitem__(self, i):
return tuple(d[i] for d in self.datasets)
def __len__(self):
return min(len(d) for d in self.datasets)
def default_loader(path):
return Image.open(path)
class CityDataSet(data.Dataset):
def __init__(self, root, split="train", img_transform=None, label_transform=None, test=True,
label_type=None):
self.root = root
self.split = split
# self.mean_bgr = np.array([104.00698793, 116.66876762, 122.67891434])
self.files = collections.defaultdict(list)
self.img_transform = img_transform
self.label_transform = label_transform
self.h_flip = HorizontalFlip()
self.v_flip = VerticalFlip()
self.test = test
data_dir = root
# for split in ["train", "trainval", "val"]:
imgsets_dir = osp.join(data_dir, "leftImg8bit/%s.txt" % split)
with open(imgsets_dir) as imgset_file:
for name in imgset_file:
name = name.strip()
img_file = osp.join(data_dir, "leftImg8bit/%s" % name)
if label_type == "label16":
name = name.replace('leftImg8bit', 'gtFine_label16IDs')
else:
name = name.replace('leftImg8bit', 'gtFine_labelTrainIds')
label_file = osp.join(data_dir, "gtFine/%s" % name)
self.files[split].append({
"img": img_file,
"label": label_file
})
def __len__(self):
return len(self.files[self.split])
def __getitem__(self, index):
datafiles = self.files[self.split][index]
img_file = datafiles["img"]
img = Image.open(img_file).convert('RGB')
label_file = datafiles["label"]
label = Image.open(label_file).convert("P")
if self.img_transform:
img = self.img_transform(img)
if self.label_transform:
label = self.label_transform(label)
if self.test:
return img, label, img_file
return img, label
class GTADataSet(data.Dataset):
def __init__(self, root, split="images", img_transform=None, label_transform=None,
test=False, input_ch=3):
# Note; split "train" and "images" are SAME!!!
assert split in ["images", "test", "train"]
assert input_ch in [1, 3, 4]
self.input_ch = input_ch
self.root = root
self.split = split
# self.mean_bgr = np.array([104.00698793, 116.66876762, 122.67891434])
self.files = collections.defaultdict(list)
self.img_transform = img_transform
self.label_transform = label_transform
self.h_flip = HorizontalFlip()
self.v_flip = VerticalFlip()
self.test = test
data_dir = root
imgsets_dir = osp.join(data_dir, "%s.txt" % split)
with open(imgsets_dir) as imgset_file:
for name in imgset_file:
name = name.strip()
img_file = osp.join(data_dir, "%s" % name)
# name = name.replace('leftImg8bit','gtFine_labelTrainIds')
label_file = osp.join(data_dir, "%s" % name.replace('images', 'labels_gt'))
self.files[split].append({
"img": img_file,
"label": label_file
})
def __len__(self):
return len(self.files[self.split])
def __getitem__(self, index):
datafiles = self.files[self.split][index]
img_file = datafiles["img"]
img = Image.open(img_file).convert('RGB')
np3ch = np.array(img)
if self.input_ch == 1:
img = ImageOps.grayscale(img)
elif self.input_ch == 4:
extended_np3ch = np.concatenate([np3ch, np3ch[:, :, 0:1]], axis=2)
img = Image.fromarray(np.uint8(extended_np3ch))
label_file = datafiles["label"]
label = Image.open(label_file).convert("P")
if self.img_transform:
img = self.img_transform(img)
if self.label_transform:
label = self.label_transform(label)
if self.test:
return img, label, img_file
return img, label
class SynthiaDataSet(data.Dataset):
def __init__(self, root, split="all", img_transform=None, label_transform=None,
test=False, input_ch=3):
# TODO this does not support "split" parameter
assert input_ch in [1, 3, 4]
self.input_ch = input_ch
self.root = root
self.split = split
self.files = collections.defaultdict(list)
self.img_transform = img_transform
self.label_transform = label_transform
self.test = test
rgb_dir = osp.join(root, "RGB")
gt_dir = osp.join(root, "GT", "LABELS16")
rgb_fn_list = glob.glob(osp.join(rgb_dir, "*.png"))
gt_fn_list = glob.glob(osp.join(gt_dir, "*.png"))
for rgb_fn, gt_fn in zip(rgb_fn_list, gt_fn_list):
self.files[split].append({
"rgb": rgb_fn,
"label": gt_fn
})
def __len__(self):
return len(self.files[self.split])
def __getitem__(self, index):
datafiles = self.files[self.split][index]
img_file = datafiles["rgb"]
img = Image.open(img_file).convert('RGB')
label_file = datafiles["label"]
label = Image.open(label_file).convert("P")
if self.img_transform:
img = self.img_transform(img)
if self.label_transform:
label = self.label_transform(label)
if self.test:
return img, label, img_file
return img, label
class TestDataSet(data.Dataset):
def __init__(self, root, split="train", img_transform=None, label_transform=None, test=True, input_ch=3):
assert input_ch == 3
self.root = root
self.split = split
# self.mean_bgr = np.array([104.00698793, 116.66876762, 122.67891434])
self.files = collections.defaultdict(list)
self.img_transform = img_transform
self.label_transform = label_transform
self.h_flip = HorizontalFlip()
self.v_flip = VerticalFlip()
self.test = test
data_dir = root
# for split in ["train", "trainval", "val"]:
imgsets_dir = os.listdir(data_dir)
for name in imgsets_dir:
img_file = osp.join(data_dir, "%s" % name)
self.files[split].append({
"img": img_file,
})
def __len__(self):
return len(self.files[self.split])
def __getitem__(self, index):
datafiles = self.files[self.split][index]
img_file = datafiles["img"]
img = Image.open(img_file).convert('RGB')
if self.img_transform:
img = self.img_transform(img)
if self.test:
return img, 'hoge', img_file
else:
return img, img
def get_dataset(dataset_name, split, img_transform, label_transform, test, input_ch=3):
assert dataset_name in ["gta", "city", "test", "city16", "synthia"]
name2obj = {
"gta": GTADataSet,
"city": CityDataSet,
"city16": CityDataSet,
"synthia": SynthiaDataSet,
}
##Note fill in the blank below !! "gta....fill the directory over images folder.
name2root = {
"gta": "", ## Fill the directory over images folder. put train.txt, val.txt in this folder
"city": "", ## ex, ./www.cityscapes-dataset.com/file-handling
"city16": "", ## Same as city
"synthia": "", ## synthia/RAND_CITYSCAPES",
}
dataset_obj = name2obj[dataset_name]
root = name2root[dataset_name]
if dataset_name == "city16":
return dataset_obj(root=root, split=split, img_transform=img_transform, label_transform=label_transform,
test=test, input_ch=input_ch, label_type="label16")
return dataset_obj(root=root, split=split, img_transform=img_transform, label_transform=label_transform,
test=test, input_ch=input_ch)
def check_src_tgt_ok(src_dataset_name, tgt_dataset_name):
if src_dataset_name == "synthia" and not tgt_dataset_name == "city16":
raise AssertionError("you must use synthia-city16 pair")
elif src_dataset_name == "city16" and not tgt_dataset_name == "synthia":
raise AssertionError("you must use synthia-city16 pair")
def get_n_class(src_dataset_name):
if src_dataset_name in ["synthia", "city16"]:
return 16
elif src_dataset_name in ["gta", "city", "test"]:
return 20
else:
raise NotImplementedError("You have to define the class of %s dataset" % src_dataset_name)