-
Notifications
You must be signed in to change notification settings - Fork 1
/
Copy pathcityscapes_mix.py
373 lines (294 loc) · 15.3 KB
/
cityscapes_mix.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
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
import json
import os
from collections import namedtuple
import torch
import torch.utils.data as data
from PIL import Image
import numpy as np
from skimage.morphology import erosion, dilation,binary_erosion, opening, closing, white_tophat, reconstruction, area_opening
from skimage.morphology import black_tophat, skeletonize, convex_hull_image,extrema
from skimage.morphology import square, diamond, octagon, rectangle, star, disk, label
from skimage.segmentation import watershed
from skimage import io, color
class Cityscapes_mix(data.Dataset):
"""Cityscapes <http://www.cityscapes-dataset.com/> Dataset.
**Parameters:**
- **root** (string): Root directory of dataset where directory 'leftImg8bit' and 'gtFine' or 'gtCoarse' are located.
- **split** (string, optional): The image split to use, 'train', 'test' or 'val' if mode="gtFine" otherwise 'train', 'train_extra' or 'val'
- **mode** (string, optional): The quality mode to use, 'gtFine' or 'gtCoarse' or 'color'. Can also be a list to output a tuple with all specified target types.
- **transform** (callable, optional): A function/transform that takes in a PIL image and returns a transformed version. E.g, ``transforms.RandomCrop``
- **target_transform** (callable, optional): A function/transform that takes in the target and transforms it.
"""
# Based on https://github.com/mcordts/cityscapesScripts
CityscapesClass = namedtuple('CityscapesClass', ['name', 'id', 'train_id', 'category', 'category_id',
'has_instances', 'ignore_in_eval', 'color'])
classes = [
CityscapesClass('unlabeled', 0, 255, 'void', 0, False, True, (0, 0, 0)),
CityscapesClass('ego vehicle', 1, 255, 'void', 0, False, True, (0, 0, 0)),
CityscapesClass('rectification border', 2, 255, 'void', 0, False, True, (0, 0, 0)),
CityscapesClass('out of roi', 3, 255, 'void', 0, False, True, (0, 0, 0)),
CityscapesClass('static', 4, 255, 'void', 0, False, True, (0, 0, 0)),
CityscapesClass('dynamic', 5, 255, 'void', 0, False, True, (111, 74, 0)),
CityscapesClass('ground', 6, 255, 'void', 0, False, True, (81, 0, 81)),
CityscapesClass('road', 7, 0, 'flat', 1, False, False, (128, 64, 128)),
CityscapesClass('sidewalk', 8, 1, 'flat', 1, False, False, (244, 35, 232)),
CityscapesClass('parking', 9, 255, 'flat', 1, False, True, (250, 170, 160)),
CityscapesClass('rail track', 10, 255, 'flat', 1, False, True, (230, 150, 140)),
CityscapesClass('building', 11, 2, 'construction', 2, False, False, (70, 70, 70)),
CityscapesClass('wall', 12, 3, 'construction', 2, False, False, (102, 102, 156)),
CityscapesClass('fence', 13, 4, 'construction', 2, False, False, (190, 153, 153)),
CityscapesClass('guard rail', 14, 255, 'construction', 2, False, True, (180, 165, 180)),
CityscapesClass('bridge', 15, 255, 'construction', 2, False, True, (150, 100, 100)),
CityscapesClass('tunnel', 16, 255, 'construction', 2, False, True, (150, 120, 90)),
CityscapesClass('pole', 17, 5, 'object', 3, False, False, (153, 153, 153)),
CityscapesClass('polegroup', 18, 255, 'object', 3, False, True, (153, 153, 153)),
CityscapesClass('traffic light', 19, 6, 'object', 3, False, False, (250, 170, 30)),
CityscapesClass('traffic sign', 20, 7, 'object', 3, False, False, (220, 220, 0)),
CityscapesClass('vegetation', 21, 8, 'nature', 4, False, False, (107, 142, 35)),
CityscapesClass('terrain', 22, 9, 'nature', 4, False, False, (152, 251, 152)),
CityscapesClass('sky', 23, 10, 'sky', 5, False, False, (70, 130, 180)),
CityscapesClass('person', 24, 11, 'human', 6, True, False, (220, 20, 60)),
CityscapesClass('rider', 25, 12, 'human', 6, True, False, (255, 0, 0)),
CityscapesClass('car', 26, 13, 'vehicle', 7, True, False, (0, 0, 142)),
CityscapesClass('truck', 27, 14, 'vehicle', 7, True, False, (0, 0, 70)),
CityscapesClass('bus', 28, 15, 'vehicle', 7, True, False, (0, 60, 100)),
CityscapesClass('caravan', 29, 255, 'vehicle', 7, True, True, (0, 0, 90)),
CityscapesClass('trailer', 30, 255, 'vehicle', 7, True, True, (0, 0, 110)),
CityscapesClass('train', 31, 16, 'vehicle', 7, True, False, (0, 80, 100)),
CityscapesClass('motorcycle', 32, 17, 'vehicle', 7, True, False, (0, 0, 230)),
CityscapesClass('bicycle', 33, 18, 'vehicle', 7, True, False, (119, 11, 32)),
CityscapesClass('license plate', -1, 255, 'vehicle', 7, False, True, (0, 0, 142)),
]
train_id_to_color = [c.color for c in classes if (c.train_id != -1 and c.train_id != 255)]
train_id_to_color.append([0, 0, 0])
train_id_to_color = np.array(train_id_to_color)
id_to_train_id = np.array([c.train_id for c in classes])
#train_id_to_color = [(0, 0, 0), (128, 64, 128), (70, 70, 70), (153, 153, 153), (107, 142, 35),
# (70, 130, 180), (220, 20, 60), (0, 0, 142)]
#train_id_to_color = np.array(train_id_to_color)
#id_to_train_id = np.array([c.category_id for c in classes], dtype='uint8') - 1
def __init__(self, root, split='train', mode='fine', target_type='semantic', transform1=None, transform2=None,watershed=False,watercutout=0.3,nb_markers=200):
self.root = os.path.expanduser(root)
self.mode = 'gtFine'
self.target_type = target_type
self.images_dir = os.path.join(self.root, 'leftImg8bit', split)
self.targets_dir = os.path.join(self.root, self.mode, split)
self.transform1 = transform1
self.transform2 = transform2
self.split = split
self.images = []
self.targets = []
self.watercutout=watercutout
self.nb_markers = nb_markers
self.watershed = watershed
if self.watershed: self.watershed_mask = Cutoutwatershed_cityscape(self.watercutout, self.nb_markers)
if split not in ['train', 'test', 'val']:
raise ValueError('Invalid split for mode! Please use split="train", split="test"'
' or split="val"')
if not os.path.isdir(self.images_dir) or not os.path.isdir(self.targets_dir):
raise RuntimeError('Dataset not found or incomplete. Please make sure all required folders for the'
' specified "split" and "mode" are inside the "root" directory')
for city in os.listdir(self.images_dir):
img_dir = os.path.join(self.images_dir, city)
target_dir = os.path.join(self.targets_dir, city)
for file_name in os.listdir(img_dir):
self.images.append(os.path.join(img_dir, file_name))
target_name = '{}_{}'.format(file_name.split('_leftImg8bit')[0],
self._get_target_suffix(self.mode, self.target_type))
self.targets.append(os.path.join(target_dir, target_name))
@classmethod
def encode_target(cls, target):
return cls.id_to_train_id[np.array(target)]
@classmethod
def decode_target(cls, target):
target[target == 255] = 19
#target = target.astype('uint8') + 1
return cls.train_id_to_color[target]
def __getitem__(self, index):
"""
Args:
index (int): Index
Returns:
tuple: (image, target) where target is a tuple of all target types if target_type is a list with more
than one item. Otherwise target is a json object if target_type="polygon", else the image segmentation.
"""
image = Image.open(self.images[index]).convert('RGB')
target = Image.open(self.targets[index])
if self.transform1:
image, target = self.transform1(image, target)
if self.watershed:
mask = torch.from_numpy(self.watershed_mask(image))
if self.transform2:
image, target = self.transform2(image, target)
target = self.encode_target(target)
if self.watershed : return image, target, mask
else: return image, target
def __len__(self):
return len(self.images)
def _load_json(self, path):
with open(path, 'r') as file:
data = json.load(file)
return data
def _get_target_suffix(self, mode, target_type):
if target_type == 'instance':
return '{}_instanceIds.png'.format(mode)
elif target_type == 'semantic':
return '{}_labelIds.png'.format(mode)
elif target_type == 'color':
return '{}_color.png'.format(mode)
elif target_type == 'polygon':
return '{}_polygons.json'.format(mode)
elif target_type == 'depth':
return '{}_disparity.png'.format(mode)
se1_0 = np.array([[ 0, 1, 0],
[ 0, 1, 0],
[ 0, 1, 0]], dtype=np.uint8)
se2_0 = np.array([[ 0, 0, 0],
[ 1, 1, 1],
[0, 0, 0]], dtype=np.uint8)
se3_0 = np.array([[ 0, 0, 1],
[0, 1, 0],
[1, 0, 0]], dtype=np.uint8)
se4_0 = np.array([[1, 0, 0],
[0, 1, 0],
[0, 0, 1]], dtype=np.uint8)
se1_1 = np.array([[0, 0, 1, 0, 0],
[0, 0, 1, 0, 0],
[0, 0, 1, 0, 0],
[0, 0, 1, 0, 0],
[0, 0, 1, 0, 0]], dtype=np.uint8)
se2_1 = np.array([[0, 0, 0, 0, 0],
[0, 0, 0, 0, 0],
[1, 1, 1, 1, 1],
[0, 0, 0, 0, 0],
[0, 0, 0, 0, 0]], dtype=np.uint8)
se3_1 = np.array([[0, 0, 0, 0, 1],
[0, 0, 0, 1, 0],
[0, 0, 1, 0, 0],
[0, 1, 0, 0, 0],
[1, 0, 0, 0, 0]], dtype=np.uint8)
se4_1 = np.array([[1, 0, 0, 0, 0],
[0, 1, 0, 0, 0],
[0, 0, 1, 0, 0],
[0, 0, 0, 1, 0],
[0, 0, 0, 0, 1]], dtype=np.uint8)
se1_2 = np.array([[0,0,0, 1, 0, 0, 0],
[0, 0, 0, 1, 0, 0, 0],
[0, 0, 0, 1, 0,0, 0],
[0, 0, 0, 1, 0, 0, 0],
[0, 0, 0, 1, 0, 0, 0],
[0, 0, 0, 1, 0, 0, 0],
[0, 0, 0, 1, 0, 0, 0]], dtype=np.uint8)
se2_2 = np.array([[0,0,0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0,0, 0],
[1, 1, 1, 1, 1, 1, 1],
[0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0]], dtype=np.uint8)
se3_2 = np.array([[0,0,0, 0, 0, 0, 1],
[0, 0, 0, 0, 0, 1, 0],
[0, 0, 0, 0, 1,0, 0],
[0, 0, 0, 1, 0, 0, 0],
[0, 0, 1, 0, 0, 0, 0],
[0, 1, 0, 0, 0, 0, 0],
[1, 0, 0, 0, 0, 0, 0]], dtype=np.uint8)
se4_2 = np.array([[1,0,0, 0, 0, 0, 0],
[0, 1, 0, 0, 0, 0, 0],
[0, 0, 1, 0, 0,0, 0],
[0, 0, 0, 1, 0, 0, 0],
[0, 0, 0, 0, 1, 0, 0],
[0, 0, 0, 0, 0, 1, 0],
[0, 0, 0, 0, 0, 0, 1]], dtype=np.uint8)
scale0=[se1_0,se2_0,se3_0,se4_0]
scale1=[se1_1,se2_1,se3_1,se4_1]
scale2=[se1_2,se2_2,se3_2,se4_2]
def gradient(rgb):
# description
# input :image rgb
# output : contour
rgb = np.asarray(rgb).astype(np.uint8)
lab = color.rgb2lab(rgb)
'''print(np.amin(lab[:, :, 0]), np.amax(lab[:, :, 0]))
print(np.amin(lab[:, :, 1]), np.amax(lab[:, :, 1]))
print(np.amin(lab[:, :, 2]), np.amax(lab[:, :, 2]))'''
##Sobel operator kernels.
tensor_grad_L=np.zeros((rgb.shape[0],rgb.shape[1],4))
tensor_grad_a = np.zeros((rgb.shape[0], rgb.shape[1], 4))
tensor_grad_b = np.zeros((rgb.shape[0], rgb.shape[1], 4))
for i in range(len(scale0)):
imgGrad = dilation(lab[:,:,0], scale0[i]) - erosion(lab[:,:,0], scale0[i])
tensor_grad_L[:, :, i] = imgGrad
imgGrad = dilation(lab[:,:,1], scale0[i]) - erosion(lab[:,:,1], scale0[i])
tensor_grad_a[:, :, i] = imgGrad
imgGrad = dilation(lab[:,:,2], scale0[i]) - erosion(lab[:,:,2], scale0[i])
tensor_grad_b[:, :, i] = imgGrad
grad_L=np.mean(tensor_grad_L,axis=2)
grad_a = np.mean(tensor_grad_a, axis=2)
grad_b = np.mean(tensor_grad_b, axis=2)
grad_scale0=np.maximum(grad_L, grad_a,grad_b)
tensor_grad_L=np.zeros((rgb.shape[0],rgb.shape[1],4))
tensor_grad_a = np.zeros((rgb.shape[0], rgb.shape[1], 4))
tensor_grad_b = np.zeros((rgb.shape[0], rgb.shape[1], 4))
for i in range(len(scale1)):
imgGrad = dilation(lab[:,:,0], scale1[i]) - erosion(lab[:,:,0], scale1[i])
tensor_grad_L[:, :, i] = imgGrad
imgGrad = dilation(lab[:,:,1], scale1[i]) - erosion(lab[:,:,1], scale1[i])
tensor_grad_a[:, :, i] = imgGrad
imgGrad = dilation(lab[:,:,2], scale1[i]) - erosion(lab[:,:,2], scale1[i])
tensor_grad_b[:, :, i] = imgGrad
grad_L=np.mean(tensor_grad_L,axis=2)
grad_a = np.mean(tensor_grad_a, axis=2)
grad_b = np.mean(tensor_grad_b, axis=2)
grad_scale1=np.maximum(grad_L, grad_a,grad_b)
tensor_grad_L=np.zeros((rgb.shape[0],rgb.shape[1],4))
tensor_grad_a = np.zeros((rgb.shape[0], rgb.shape[1], 4))
tensor_grad_b = np.zeros((rgb.shape[0], rgb.shape[1], 4))
for i in range(len(scale2)):
imgGrad = dilation(lab[:,:,0], scale2[i]) - erosion(lab[:,:,0], scale2[i])
tensor_grad_L[:, :, i] = imgGrad
imgGrad = dilation(lab[:,:,1], scale2[i]) - erosion(lab[:,:,1], scale2[i])
tensor_grad_a[:, :, i] = imgGrad
imgGrad = dilation(lab[:,:,2], scale2[i]) - erosion(lab[:,:,2], scale2[i])
tensor_grad_b[:, :, i] = imgGrad
grad_L=np.mean(tensor_grad_L,axis=2)
grad_a = np.mean(tensor_grad_a, axis=2)
grad_b = np.mean(tensor_grad_b, axis=2)
grad_scale2=np.maximum(grad_L, grad_a,grad_b)
grad=(grad_scale0+grad_scale1+grad_scale2)/3
return grad
def mosaic_cutout_binary(rgb,labels_waterhed,prop):
shape = np.shape(rgb)
mask = np.ones((shape[0],shape[1])).astype(np.float16)
nb_cluster=np.amax(labels_waterhed)
perm = np.random.permutation(nb_cluster)
nb=int(nb_cluster*prop)
nb_cluster=perm[0:nb]
for i in nb_cluster:
mask[labels_waterhed == i] = 0
return mask
class Cutoutwatershed_cityscape(object):
"""
Reference : https://github.com/quark0/darts/blob/master/cnn/utils.py
"""
def __init__(self, prop,nb_markers=200):
self.prop = prop
self.nb_markers=nb_markers
def __call__(self, img):
if self.prop <= 0:
return img
img = np.asarray(img)
size=img.shape
img2 = img.copy()#np.ones((size[1],size[2],3)).astype(np.uint8)
'''r,g,b = np.split(img, 3, axis=0)
print(np.shape(r))
img2[:, :, 0] = r
img2[:, :, 1] = g
img2[:, :, 2] = b'''
img2=Image.fromarray(img2)
grad =gradient(img2)
labels_waterhed = watershed(grad, markers=200, compactness=0.001)
img=mosaic_cutout_binary(img2,labels_waterhed,self.prop)
#Image.fromarray(segments_watershed).show('img_mosaic')
#print(img,img.shape)
return img