├── README.md
├── data
└── tfrecord
│ ├── ReadMe
│ ├── sample_images_list.txt
│ ├── test.txt
│ ├── train.txt
│ └── val.txt
├── output
├── 1.jpg
├── 2.jpg
├── 2007_000027.jpg
├── 2007_000121.jpg
├── 2007_000243.jpg
├── 2007_000346.jpg
├── 2007_000364.jpg
├── 2007_000452.jpg
├── 2007_000464.jpg
├── 2007_000529.jpg
├── 3.jpg
├── 4.jpg
├── 5.jpg
├── UNET.png
├── model1.png
├── model2.png
├── test.jpg
├── u=3476430323,4263663876&fm=11&gp=0.jpg
├── u=3893090294,1830313637&fm=11&gp=0.jpg
└── w475_h331_9a5169d0369e4e1496d1cdfabb1ded85.jpg
├── picture
├── 1.jpg
├── 2.jpg
├── 2007_000027.jpg
├── 2007_000121.jpg
├── 2007_000243.jpg
├── 2007_000346.jpg
├── 2007_000364.jpg
├── 2007_000452.jpg
├── 2007_000464.jpg
├── 2007_000529.jpg
├── 3.jpg
├── 4.jpg
├── 5.jpg
├── test.jpg
├── u=3476430323,4263663876&fm=11&gp=0.jpg
├── u=3893090294,1830313637&fm=11&gp=0.jpg
└── w475_h331_9a5169d0369e4e1496d1cdfabb1ded85.jpg
├── test.py
├── tfrecord.py
├── train.py
└── utils
├── __init__.py
├── __pycache__
├── __init__.cpython-35.pyc
├── __init__.cpython-36.pyc
├── config.cpython-35.pyc
├── config.cpython-36.pyc
├── dataset_util.cpython-35.pyc
├── deeplab1_model.cpython-35.pyc
├── deeplab_model.cpython-35.pyc
├── deeplab_model.cpython-36.pyc
├── deeplab_model1.cpython-35.pyc
├── model.cpython-35.pyc
├── preprocessing.cpython-35.pyc
└── preprocessing.cpython-36.pyc
├── config.py
├── dataset_util.py
├── deeplab_model.py
└── preprocessing.py
/README.md:
--------------------------------------------------------------------------------
1 | # tensorflow-deeplab_v3_plus
2 | 参考[rishizek](https://github.com/rishizek/tensorflow-deeplab-v3-plus)的代码进行中文注释，并按照自己风格重新编写代码，对ASPP加入里BN层，支持摄像头。
3 | ## deeplab_v3_plus简介
4 | 图像分割是主要功能是将输入图片的每个像素都分好类别，也相当于分类过程。举例来说就是将大小为[h,w,c]的图像输出成[h,w,1]，每个像素值代表一个类别。
5 | deeplab_v3+可以参考论文[Encoder-Decoder with Atrous Separable Convolution for Semantic Image Segmentation](https://arxiv.org/pdf/1802.02611.pdf)。它的结构图如下：
6 | 
7 | 
8 | 下面对模型进行简要分析
9 | 该模型属于encoder-decoder模型，encoder-decoder常用于自然语言处理中，在图像分割中[U-net](https://arxiv.org/pdf/1505.04597.pdf)也是十分典型的encoder-decoder模型，大体结构如下：
10 | 
11 | 就是将图片通过卷积尺寸变小再通过上采样将尺寸还原。
12 | deeplab_v3+是将encoder-decoder和ASPP相结合，encoder-decoder会获取更多边界信息，ASPP可获取更多特征信息。encoder网络使用resnet101或 Xception,本代码中使用的是resnet101。
13 | 采用预训练的resnet的某一节点A来获取图像信息，再加入到ASPP中。ASPP就是不同膨胀率的空洞卷积和全局池化上采样后的输出concat在一起，作为encoder输出部分。
14 | 空洞卷积可以理解为一个大卷积中间权重值都为0,举例说明，一个3x3的卷积，如果膨胀率是1就是正常卷积，如果膨胀率是2,就是空洞卷积，相当于把3x3的卷积每个值的右方和下方加一行或列都置0。变换之后的空洞矩阵大小变为6x6。空洞矩阵论文中说可以提取更密集的特征，捕获多尺度信息，相比于卷积和池化会减少信息丢失。全局池化就是将输入[h,w,c]池化成[1,1,c]。
15 | decoder部分选取resnet中A节点之前的B节点，再将encoder的输出上采样成B的大小然后concat，做一些卷积和上采样就得到最终输出。
16 | 由于可以看成分类问题，该模型的损失函数也是交叉熵函数。模型具体实现可以参考代码
17 | ## 模型训练
18 | ### 环境要求
19 | ubuntu=16.04
20 | tensorflow=1.4.1
21 | opencv=3.4.1
22 | windows下可以进行测试
23 | ### 下载数据集
24 | 将[VOC](http://host.robots.ox.ac.uk/pascal/VOC/voc2012/VOCtrainval_11-May-2012.tar)解压到data目录下,下载DrSleep提供的[SegmentationClassAug文件](https://www.dropbox.com/s/oeu149j8qtbs1x0/SegmentationClassAug.zip?dl=0)解压，如果访问不了可以尝试[这个网址](https://www.jianguoyun.com/p/DQVKi9QQv7mYBxjfsIwB)里面有_MACOSX和SegmentationClassAug两个文件夹，将里面的SegmentationClassAug放置到./data//VOCdevkit/VOC2012/下，里面的图片是shape为[h,w,1]每一个像素值都对应类别的label。
25 | 将[restnet预训练数据](https://www.jianguoyun.com/p/DZeHNHwQv7mYBxiBsIwB)解压放置在该模型的根目录下。
26 | 如果需要模型预训练数据可以将我训练的[权重数据](https://share.weiyun.com/5lGifzi)解压，如果下载过慢可以尝试[这个网址](https://www.jianguoyun.com/p/DYz_5HkQ9_yeBxilsYwB),将里面的ckpt等文件放置到./model下。
27 | ### 代码介绍
28 | data放置VOC数据和数据处理生成的record文件和分化数据的txt文件
29 | model放置训练生成的模型和graph
30 | output放置测试图片生成的分割图像
31 | picture放置测试用例,我的来源于百度图片
32 | utils包含配置文件config.py,数据处理文件dataset_util.py,preprocessing.py和模型文件deeplab_model.py
33 | test.py是测试文件支持摄像头
34 | tfrecord.py是将处理完的数据生成record文件
35 | train.py是训练文件
36 | ### 运行
37 | 手动配置config.py的信息或选择默认
38 | 若要训练：
39 | 运行python tfrecord.py生成record文件
40 | 运行python train.py训练。 对于计算机gpu选择需修改[这里代码](https://github.com/LeslieZhoa/tensorflow-deeplab_v3_plus/blob/master/train.py#L38)
41 | 若要测试：
42 | 运行python test.py
43 | ## 一些疑问
44 | 我的电脑配置是1080Ti但总是运行运行就溢出，我尝试用tf.contrib.distribute.mirroredstrategy多gpu并行，但tensorflow版本要1.8,当我更新完，发现input_fn要是data格式，我失败了。
45 | 如果有并行gpu的建议或者代码的指正请给我留言
46 | ## 结果展示
47 | 
48 | 
49 | 
50 |
51 |
--------------------------------------------------------------------------------
/data/tfrecord/ReadMe:
--------------------------------------------------------------------------------
1 | 放置record数据
2 |
3 |
--------------------------------------------------------------------------------
/data/tfrecord/sample_images_list.txt:
--------------------------------------------------------------------------------
1 | 2007_000033.jpg
2 | 2007_000042.jpg
3 | 2007_000061.jpg
4 | 2007_000123.jpg
5 | 2007_000129.jpg
6 | 2007_000175.jpg
7 | 2007_000187.jpg
8 | 2007_000323.jpg
--------------------------------------------------------------------------------
/data/tfrecord/test.txt:
--------------------------------------------------------------------------------
1 | 2008_000006
2 | 2008_000011
3 | 2008_000012
4 | 2008_000018
5 | 2008_000024
6 | 2008_000030
7 | 2008_000031
8 | 2008_000046
9 | 2008_000047
10 | 2008_000048
11 | 2008_000057
12 | 2008_000058
13 | 2008_000068
14 | 2008_000072
15 | 2008_000079
16 | 2008_000081
17 | 2008_000083
18 | 2008_000088
19 | 2008_000094
20 | 2008_000101
21 | 2008_000104
22 | 2008_000106
23 | 2008_000108
24 | 2008_000110
25 | 2008_000111
26 | 2008_000126
27 | 2008_000127
28 | 2008_000129
29 | 2008_000130
30 | 2008_000135
31 | 2008_000150
32 | 2008_000152
33 | 2008_000156
34 | 2008_000159
35 | 2008_000160
36 | 2008_000161
37 | 2008_000166
38 | 2008_000167
39 | 2008_000168
40 | 2008_000169
41 | 2008_000171
42 | 2008_000175
43 | 2008_000178
44 | 2008_000186
45 | 2008_000198
46 | 2008_000206
47 | 2008_000208
48 | 2008_000209
49 | 2008_000211
50 | 2008_000220
51 | 2008_000224
52 | 2008_000230
53 | 2008_000240
54 | 2008_000248
55 | 2008_000249
56 | 2008_000250
57 | 2008_000256
58 | 2008_000279
59 | 2008_000282
60 | 2008_000285
61 | 2008_000286
62 | 2008_000296
63 | 2008_000300
64 | 2008_000322
65 | 2008_000324
66 | 2008_000337
67 | 2008_000366
68 | 2008_000369
69 | 2008_000377
70 | 2008_000384
71 | 2008_000390
72 | 2008_000404
73 | 2008_000411
74 | 2008_000434
75 | 2008_000440
76 | 2008_000460
77 | 2008_000467
78 | 2008_000478
79 | 2008_000485
80 | 2008_000487
81 | 2008_000490
82 | 2008_000503
83 | 2008_000504
84 | 2008_000507
85 | 2008_000513
86 | 2008_000523
87 | 2008_000529
88 | 2008_000556
89 | 2008_000565
90 | 2008_000580
91 | 2008_000590
92 | 2008_000596
93 | 2008_000597
94 | 2008_000600
95 | 2008_000603
96 | 2008_000604
97 | 2008_000612
98 | 2008_000617
99 | 2008_000621
100 | 2008_000627
101 | 2008_000633
102 | 2008_000643
103 | 2008_000644
104 | 2008_000649
105 | 2008_000651
106 | 2008_000664
107 | 2008_000665
108 | 2008_000680
109 | 2008_000681
110 | 2008_000684
111 | 2008_000685
112 | 2008_000688
113 | 2008_000693
114 | 2008_000698
115 | 2008_000707
116 | 2008_000709
117 | 2008_000712
118 | 2008_000747
119 | 2008_000751
120 | 2008_000754
121 | 2008_000762
122 | 2008_000767
123 | 2008_000768
124 | 2008_000773
125 | 2008_000774
126 | 2008_000779
127 | 2008_000797
128 | 2008_000813
129 | 2008_000816
130 | 2008_000846
131 | 2008_000866
132 | 2008_000871
133 | 2008_000872
134 | 2008_000891
135 | 2008_000892
136 | 2008_000894
137 | 2008_000896
138 | 2008_000898
139 | 2008_000909
140 | 2008_000913
141 | 2008_000920
142 | 2008_000933
143 | 2008_000935
144 | 2008_000937
145 | 2008_000938
146 | 2008_000954
147 | 2008_000958
148 | 2008_000963
149 | 2008_000967
150 | 2008_000974
151 | 2008_000986
152 | 2008_000994
153 | 2008_000995
154 | 2008_001008
155 | 2008_001010
156 | 2008_001014
157 | 2008_001016
158 | 2008_001025
159 | 2008_001029
160 | 2008_001037
161 | 2008_001059
162 | 2008_001061
163 | 2008_001072
164 | 2008_001124
165 | 2008_001126
166 | 2008_001131
167 | 2008_001138
168 | 2008_001144
169 | 2008_001151
170 | 2008_001156
171 | 2008_001179
172 | 2008_001181
173 | 2008_001184
174 | 2008_001186
175 | 2008_001197
176 | 2008_001207
177 | 2008_001212
178 | 2008_001233
179 | 2008_001234
180 | 2008_001258
181 | 2008_001268
182 | 2008_001279
183 | 2008_001281
184 | 2008_001288
185 | 2008_001291
186 | 2008_001298
187 | 2008_001309
188 | 2008_001315
189 | 2008_001316
190 | 2008_001319
191 | 2008_001327
192 | 2008_001328
193 | 2008_001332
194 | 2008_001341
195 | 2008_001347
196 | 2008_001355
197 | 2008_001378
198 | 2008_001386
199 | 2008_001400
200 | 2008_001409
201 | 2008_001411
202 | 2008_001416
203 | 2008_001418
204 | 2008_001435
205 | 2008_001459
206 | 2008_001469
207 | 2008_001474
208 | 2008_001477
209 | 2008_001483
210 | 2008_001484
211 | 2008_001485
212 | 2008_001496
213 | 2008_001507
214 | 2008_001511
215 | 2008_001519
216 | 2008_001557
217 | 2008_001567
218 | 2008_001570
219 | 2008_001571
220 | 2008_001572
221 | 2008_001579
222 | 2008_001587
223 | 2008_001608
224 | 2008_001611
225 | 2008_001614
226 | 2008_001621
227 | 2008_001639
228 | 2008_001658
229 | 2008_001678
230 | 2008_001700
231 | 2008_001713
232 | 2008_001720
233 | 2008_001755
234 | 2008_001779
235 | 2008_001785
236 | 2008_001793
237 | 2008_001794
238 | 2008_001803
239 | 2008_001818
240 | 2008_001848
241 | 2008_001855
242 | 2008_001857
243 | 2008_001861
244 | 2008_001875
245 | 2008_001878
246 | 2008_001886
247 | 2008_001897
248 | 2008_001916
249 | 2008_001925
250 | 2008_001949
251 | 2008_001953
252 | 2008_001972
253 | 2008_001999
254 | 2008_002027
255 | 2008_002040
256 | 2008_002057
257 | 2008_002070
258 | 2008_002075
259 | 2008_002095
260 | 2008_002104
261 | 2008_002105
262 | 2008_002106
263 | 2008_002136
264 | 2008_002137
265 | 2008_002147
266 | 2008_002149
267 | 2008_002163
268 | 2008_002173
269 | 2008_002174
270 | 2008_002184
271 | 2008_002186
272 | 2008_002188
273 | 2008_002190
274 | 2008_002203
275 | 2008_002211
276 | 2008_002217
277 | 2008_002228
278 | 2008_002233
279 | 2008_002246
280 | 2008_002257
281 | 2008_002261
282 | 2008_002285
283 | 2008_002287
284 | 2008_002295
285 | 2008_002303
286 | 2008_002306
287 | 2008_002309
288 | 2008_002310
289 | 2008_002318
290 | 2008_002320
291 | 2008_002332
292 | 2008_002337
293 | 2008_002345
294 | 2008_002348
295 | 2008_002352
296 | 2008_002360
297 | 2008_002381
298 | 2008_002387
299 | 2008_002388
300 | 2008_002393
301 | 2008_002406
302 | 2008_002440
303 | 2008_002455
304 | 2008_002460
305 | 2008_002462
306 | 2008_002480
307 | 2008_002518
308 | 2008_002525
309 | 2008_002535
310 | 2008_002544
311 | 2008_002553
312 | 2008_002569
313 | 2008_002572
314 | 2008_002587
315 | 2008_002635
316 | 2008_002655
317 | 2008_002695
318 | 2008_002702
319 | 2008_002706
320 | 2008_002707
321 | 2008_002722
322 | 2008_002745
323 | 2008_002757
324 | 2008_002779
325 | 2008_002805
326 | 2008_002871
327 | 2008_002895
328 | 2008_002905
329 | 2008_002923
330 | 2008_002927
331 | 2008_002939
332 | 2008_002941
333 | 2008_002962
334 | 2008_002975
335 | 2008_003000
336 | 2008_003031
337 | 2008_003038
338 | 2008_003042
339 | 2008_003069
340 | 2008_003070
341 | 2008_003115
342 | 2008_003116
343 | 2008_003130
344 | 2008_003137
345 | 2008_003138
346 | 2008_003139
347 | 2008_003165
348 | 2008_003171
349 | 2008_003176
350 | 2008_003192
351 | 2008_003194
352 | 2008_003195
353 | 2008_003198
354 | 2008_003227
355 | 2008_003247
356 | 2008_003262
357 | 2008_003298
358 | 2008_003299
359 | 2008_003307
360 | 2008_003337
361 | 2008_003353
362 | 2008_003355
363 | 2008_003363
364 | 2008_003383
365 | 2008_003389
366 | 2008_003392
367 | 2008_003399
368 | 2008_003436
369 | 2008_003457
370 | 2008_003465
371 | 2008_003481
372 | 2008_003539
373 | 2008_003548
374 | 2008_003550
375 | 2008_003567
376 | 2008_003568
377 | 2008_003606
378 | 2008_003615
379 | 2008_003654
380 | 2008_003670
381 | 2008_003700
382 | 2008_003705
383 | 2008_003727
384 | 2008_003731
385 | 2008_003734
386 | 2008_003760
387 | 2008_003804
388 | 2008_003807
389 | 2008_003810
390 | 2008_003822
391 | 2008_003833
392 | 2008_003877
393 | 2008_003879
394 | 2008_003895
395 | 2008_003901
396 | 2008_003903
397 | 2008_003911
398 | 2008_003919
399 | 2008_003927
400 | 2008_003937
401 | 2008_003946
402 | 2008_003950
403 | 2008_003955
404 | 2008_003981
405 | 2008_003991
406 | 2008_004009
407 | 2008_004039
408 | 2008_004052
409 | 2008_004063
410 | 2008_004070
411 | 2008_004078
412 | 2008_004104
413 | 2008_004139
414 | 2008_004177
415 | 2008_004181
416 | 2008_004200
417 | 2008_004219
418 | 2008_004236
419 | 2008_004250
420 | 2008_004266
421 | 2008_004299
422 | 2008_004320
423 | 2008_004334
424 | 2008_004343
425 | 2008_004349
426 | 2008_004366
427 | 2008_004386
428 | 2008_004401
429 | 2008_004423
430 | 2008_004448
431 | 2008_004481
432 | 2008_004516
433 | 2008_004536
434 | 2008_004582
435 | 2008_004609
436 | 2008_004638
437 | 2008_004642
438 | 2008_004644
439 | 2008_004669
440 | 2008_004673
441 | 2008_004691
442 | 2008_004693
443 | 2008_004709
444 | 2008_004715
445 | 2008_004757
446 | 2008_004775
447 | 2008_004782
448 | 2008_004785
449 | 2008_004798
450 | 2008_004848
451 | 2008_004861
452 | 2008_004870
453 | 2008_004877
454 | 2008_004884
455 | 2008_004891
456 | 2008_004901
457 | 2008_004919
458 | 2008_005058
459 | 2008_005069
460 | 2008_005086
461 | 2008_005087
462 | 2008_005112
463 | 2008_005113
464 | 2008_005118
465 | 2008_005128
466 | 2008_005129
467 | 2008_005153
468 | 2008_005161
469 | 2008_005162
470 | 2008_005165
471 | 2008_005187
472 | 2008_005227
473 | 2008_005308
474 | 2008_005318
475 | 2008_005320
476 | 2008_005351
477 | 2008_005372
478 | 2008_005383
479 | 2008_005391
480 | 2008_005407
481 | 2008_005420
482 | 2008_005440
483 | 2008_005487
484 | 2008_005493
485 | 2008_005520
486 | 2008_005551
487 | 2008_005556
488 | 2008_005576
489 | 2008_005578
490 | 2008_005594
491 | 2008_005619
492 | 2008_005629
493 | 2008_005644
494 | 2008_005645
495 | 2008_005651
496 | 2008_005661
497 | 2008_005662
498 | 2008_005667
499 | 2008_005694
500 | 2008_005697
501 | 2008_005709
502 | 2008_005710
503 | 2008_005733
504 | 2008_005749
505 | 2008_005753
506 | 2008_005771
507 | 2008_005781
508 | 2008_005793
509 | 2008_005802
510 | 2008_005833
511 | 2008_005844
512 | 2008_005908
513 | 2008_005931
514 | 2008_005952
515 | 2008_006016
516 | 2008_006030
517 | 2008_006033
518 | 2008_006054
519 | 2008_006073
520 | 2008_006091
521 | 2008_006142
522 | 2008_006150
523 | 2008_006206
524 | 2008_006217
525 | 2008_006264
526 | 2008_006283
527 | 2008_006308
528 | 2008_006313
529 | 2008_006333
530 | 2008_006343
531 | 2008_006381
532 | 2008_006391
533 | 2008_006423
534 | 2008_006428
535 | 2008_006440
536 | 2008_006444
537 | 2008_006473
538 | 2008_006505
539 | 2008_006531
540 | 2008_006560
541 | 2008_006571
542 | 2008_006582
543 | 2008_006594
544 | 2008_006601
545 | 2008_006633
546 | 2008_006653
547 | 2008_006678
548 | 2008_006755
549 | 2008_006772
550 | 2008_006788
551 | 2008_006799
552 | 2008_006809
553 | 2008_006838
554 | 2008_006845
555 | 2008_006852
556 | 2008_006894
557 | 2008_006905
558 | 2008_006947
559 | 2008_006983
560 | 2008_007049
561 | 2008_007065
562 | 2008_007068
563 | 2008_007111
564 | 2008_007148
565 | 2008_007159
566 | 2008_007193
567 | 2008_007228
568 | 2008_007235
569 | 2008_007249
570 | 2008_007255
571 | 2008_007268
572 | 2008_007275
573 | 2008_007292
574 | 2008_007299
575 | 2008_007306
576 | 2008_007316
577 | 2008_007400
578 | 2008_007401
579 | 2008_007419
580 | 2008_007437
581 | 2008_007483
582 | 2008_007487
583 | 2008_007520
584 | 2008_007551
585 | 2008_007603
586 | 2008_007616
587 | 2008_007654
588 | 2008_007663
589 | 2008_007708
590 | 2008_007795
591 | 2008_007801
592 | 2008_007859
593 | 2008_007903
594 | 2008_007920
595 | 2008_007926
596 | 2008_008014
597 | 2008_008017
598 | 2008_008060
599 | 2008_008077
600 | 2008_008107
601 | 2008_008108
602 | 2008_008119
603 | 2008_008126
604 | 2008_008133
605 | 2008_008144
606 | 2008_008216
607 | 2008_008244
608 | 2008_008248
609 | 2008_008250
610 | 2008_008260
611 | 2008_008277
612 | 2008_008280
613 | 2008_008290
614 | 2008_008304
615 | 2008_008340
616 | 2008_008371
617 | 2008_008390
618 | 2008_008397
619 | 2008_008409
620 | 2008_008412
621 | 2008_008419
622 | 2008_008454
623 | 2008_008491
624 | 2008_008498
625 | 2008_008565
626 | 2008_008599
627 | 2008_008603
628 | 2008_008631
629 | 2008_008634
630 | 2008_008640
631 | 2008_008646
632 | 2008_008660
633 | 2008_008663
634 | 2008_008664
635 | 2008_008709
636 | 2008_008720
637 | 2008_008747
638 | 2008_008768
639 | 2009_000004
640 | 2009_000019
641 | 2009_000024
642 | 2009_000025
643 | 2009_000053
644 | 2009_000076
645 | 2009_000107
646 | 2009_000110
647 | 2009_000115
648 | 2009_000117
649 | 2009_000175
650 | 2009_000220
651 | 2009_000259
652 | 2009_000275
653 | 2009_000314
654 | 2009_000368
655 | 2009_000373
656 | 2009_000384
657 | 2009_000388
658 | 2009_000423
659 | 2009_000433
660 | 2009_000434
661 | 2009_000458
662 | 2009_000475
663 | 2009_000481
664 | 2009_000495
665 | 2009_000514
666 | 2009_000555
667 | 2009_000556
668 | 2009_000561
669 | 2009_000571
670 | 2009_000581
671 | 2009_000605
672 | 2009_000609
673 | 2009_000644
674 | 2009_000654
675 | 2009_000671
676 | 2009_000733
677 | 2009_000740
678 | 2009_000766
679 | 2009_000775
680 | 2009_000776
681 | 2009_000795
682 | 2009_000850
683 | 2009_000881
684 | 2009_000900
685 | 2009_000914
686 | 2009_000941
687 | 2009_000977
688 | 2009_000984
689 | 2009_000986
690 | 2009_001005
691 | 2009_001015
692 | 2009_001058
693 | 2009_001072
694 | 2009_001087
695 | 2009_001092
696 | 2009_001109
697 | 2009_001114
698 | 2009_001115
699 | 2009_001141
700 | 2009_001174
701 | 2009_001175
702 | 2009_001182
703 | 2009_001222
704 | 2009_001228
705 | 2009_001246
706 | 2009_001262
707 | 2009_001274
708 | 2009_001284
709 | 2009_001297
710 | 2009_001331
711 | 2009_001336
712 | 2009_001337
713 | 2009_001379
714 | 2009_001392
715 | 2009_001451
716 | 2009_001485
717 | 2009_001488
718 | 2009_001497
719 | 2009_001504
720 | 2009_001506
721 | 2009_001573
722 | 2009_001576
723 | 2009_001603
724 | 2009_001613
725 | 2009_001652
726 | 2009_001661
727 | 2009_001668
728 | 2009_001680
729 | 2009_001688
730 | 2009_001697
731 | 2009_001729
732 | 2009_001771
733 | 2009_001785
734 | 2009_001793
735 | 2009_001814
736 | 2009_001866
737 | 2009_001872
738 | 2009_001880
739 | 2009_001883
740 | 2009_001891
741 | 2009_001913
742 | 2009_001938
743 | 2009_001946
744 | 2009_001953
745 | 2009_001969
746 | 2009_001978
747 | 2009_001995
748 | 2009_002007
749 | 2009_002036
750 | 2009_002041
751 | 2009_002049
752 | 2009_002051
753 | 2009_002062
754 | 2009_002063
755 | 2009_002067
756 | 2009_002085
757 | 2009_002092
758 | 2009_002114
759 | 2009_002115
760 | 2009_002142
761 | 2009_002148
762 | 2009_002157
763 | 2009_002181
764 | 2009_002220
765 | 2009_002284
766 | 2009_002287
767 | 2009_002300
768 | 2009_002310
769 | 2009_002315
770 | 2009_002334
771 | 2009_002337
772 | 2009_002354
773 | 2009_002357
774 | 2009_002411
775 | 2009_002426
776 | 2009_002458
777 | 2009_002459
778 | 2009_002461
779 | 2009_002466
780 | 2009_002481
781 | 2009_002483
782 | 2009_002503
783 | 2009_002581
784 | 2009_002583
785 | 2009_002589
786 | 2009_002600
787 | 2009_002601
788 | 2009_002602
789 | 2009_002641
790 | 2009_002646
791 | 2009_002656
792 | 2009_002666
793 | 2009_002720
794 | 2009_002767
795 | 2009_002768
796 | 2009_002794
797 | 2009_002821
798 | 2009_002825
799 | 2009_002839
800 | 2009_002840
801 | 2009_002859
802 | 2009_002860
803 | 2009_002881
804 | 2009_002889
805 | 2009_002892
806 | 2009_002895
807 | 2009_002896
808 | 2009_002900
809 | 2009_002924
810 | 2009_002966
811 | 2009_002973
812 | 2009_002981
813 | 2009_003004
814 | 2009_003021
815 | 2009_003028
816 | 2009_003037
817 | 2009_003038
818 | 2009_003055
819 | 2009_003085
820 | 2009_003100
821 | 2009_003106
822 | 2009_003117
823 | 2009_003139
824 | 2009_003170
825 | 2009_003179
826 | 2009_003184
827 | 2009_003186
828 | 2009_003190
829 | 2009_003221
830 | 2009_003236
831 | 2009_003242
832 | 2009_003244
833 | 2009_003260
834 | 2009_003264
835 | 2009_003274
836 | 2009_003283
837 | 2009_003296
838 | 2009_003332
839 | 2009_003341
840 | 2009_003354
841 | 2009_003370
842 | 2009_003371
843 | 2009_003374
844 | 2009_003391
845 | 2009_003393
846 | 2009_003404
847 | 2009_003405
848 | 2009_003414
849 | 2009_003428
850 | 2009_003470
851 | 2009_003474
852 | 2009_003532
853 | 2009_003536
854 | 2009_003578
855 | 2009_003580
856 | 2009_003620
857 | 2009_003621
858 | 2009_003680
859 | 2009_003699
860 | 2009_003727
861 | 2009_003737
862 | 2009_003780
863 | 2009_003811
864 | 2009_003824
865 | 2009_003831
866 | 2009_003844
867 | 2009_003850
868 | 2009_003851
869 | 2009_003864
870 | 2009_003868
871 | 2009_003869
872 | 2009_003893
873 | 2009_003909
874 | 2009_003924
875 | 2009_003925
876 | 2009_003960
877 | 2009_003979
878 | 2009_003990
879 | 2009_003997
880 | 2009_004006
881 | 2009_004010
882 | 2009_004066
883 | 2009_004077
884 | 2009_004081
885 | 2009_004097
886 | 2009_004098
887 | 2009_004136
888 | 2009_004216
889 | 2009_004220
890 | 2009_004266
891 | 2009_004269
892 | 2009_004286
893 | 2009_004296
894 | 2009_004321
895 | 2009_004342
896 | 2009_004343
897 | 2009_004344
898 | 2009_004385
899 | 2009_004408
900 | 2009_004420
901 | 2009_004441
902 | 2009_004447
903 | 2009_004461
904 | 2009_004467
905 | 2009_004485
906 | 2009_004488
907 | 2009_004516
908 | 2009_004521
909 | 2009_004544
910 | 2009_004596
911 | 2009_004613
912 | 2009_004615
913 | 2009_004618
914 | 2009_004621
915 | 2009_004646
916 | 2009_004659
917 | 2009_004663
918 | 2009_004666
919 | 2009_004691
920 | 2009_004715
921 | 2009_004726
922 | 2009_004753
923 | 2009_004776
924 | 2009_004811
925 | 2009_004814
926 | 2009_004818
927 | 2009_004835
928 | 2009_004863
929 | 2009_004894
930 | 2009_004909
931 | 2009_004928
932 | 2009_004937
933 | 2009_004954
934 | 2009_004966
935 | 2009_004970
936 | 2009_004976
937 | 2009_005004
938 | 2009_005011
939 | 2009_005053
940 | 2009_005072
941 | 2009_005115
942 | 2009_005146
943 | 2009_005151
944 | 2009_005164
945 | 2009_005179
946 | 2009_005224
947 | 2009_005243
948 | 2009_005249
949 | 2009_005252
950 | 2009_005254
951 | 2009_005258
952 | 2009_005264
953 | 2009_005266
954 | 2009_005276
955 | 2009_005290
956 | 2009_005295
957 | 2010_000004
958 | 2010_000005
959 | 2010_000006
960 | 2010_000032
961 | 2010_000062
962 | 2010_000093
963 | 2010_000094
964 | 2010_000161
965 | 2010_000176
966 | 2010_000223
967 | 2010_000226
968 | 2010_000236
969 | 2010_000239
970 | 2010_000287
971 | 2010_000300
972 | 2010_000301
973 | 2010_000328
974 | 2010_000378
975 | 2010_000405
976 | 2010_000407
977 | 2010_000472
978 | 2010_000479
979 | 2010_000491
980 | 2010_000533
981 | 2010_000535
982 | 2010_000542
983 | 2010_000554
984 | 2010_000580
985 | 2010_000594
986 | 2010_000596
987 | 2010_000599
988 | 2010_000606
989 | 2010_000615
990 | 2010_000654
991 | 2010_000659
992 | 2010_000693
993 | 2010_000698
994 | 2010_000730
995 | 2010_000734
996 | 2010_000741
997 | 2010_000755
998 | 2010_000768
999 | 2010_000794
1000 | 2010_000813
1001 | 2010_000817
1002 | 2010_000834
1003 | 2010_000839
1004 | 2010_000848
1005 | 2010_000881
1006 | 2010_000888
1007 | 2010_000900
1008 | 2010_000903
1009 | 2010_000924
1010 | 2010_000946
1011 | 2010_000953
1012 | 2010_000957
1013 | 2010_000967
1014 | 2010_000992
1015 | 2010_000998
1016 | 2010_001053
1017 | 2010_001067
1018 | 2010_001114
1019 | 2010_001132
1020 | 2010_001138
1021 | 2010_001169
1022 | 2010_001171
1023 | 2010_001228
1024 | 2010_001260
1025 | 2010_001268
1026 | 2010_001280
1027 | 2010_001298
1028 | 2010_001302
1029 | 2010_001308
1030 | 2010_001324
1031 | 2010_001332
1032 | 2010_001335
1033 | 2010_001345
1034 | 2010_001346
1035 | 2010_001349
1036 | 2010_001373
1037 | 2010_001381
1038 | 2010_001392
1039 | 2010_001396
1040 | 2010_001420
1041 | 2010_001500
1042 | 2010_001506
1043 | 2010_001521
1044 | 2010_001532
1045 | 2010_001558
1046 | 2010_001598
1047 | 2010_001611
1048 | 2010_001631
1049 | 2010_001639
1050 | 2010_001651
1051 | 2010_001663
1052 | 2010_001664
1053 | 2010_001728
1054 | 2010_001778
1055 | 2010_001861
1056 | 2010_001874
1057 | 2010_001900
1058 | 2010_001905
1059 | 2010_001969
1060 | 2010_002008
1061 | 2010_002014
1062 | 2010_002049
1063 | 2010_002052
1064 | 2010_002091
1065 | 2010_002115
1066 | 2010_002119
1067 | 2010_002134
1068 | 2010_002156
1069 | 2010_002160
1070 | 2010_002186
1071 | 2010_002210
1072 | 2010_002241
1073 | 2010_002252
1074 | 2010_002258
1075 | 2010_002262
1076 | 2010_002273
1077 | 2010_002290
1078 | 2010_002292
1079 | 2010_002347
1080 | 2010_002358
1081 | 2010_002360
1082 | 2010_002367
1083 | 2010_002416
1084 | 2010_002451
1085 | 2010_002481
1086 | 2010_002490
1087 | 2010_002495
1088 | 2010_002588
1089 | 2010_002607
1090 | 2010_002609
1091 | 2010_002610
1092 | 2010_002641
1093 | 2010_002685
1094 | 2010_002699
1095 | 2010_002719
1096 | 2010_002735
1097 | 2010_002751
1098 | 2010_002804
1099 | 2010_002835
1100 | 2010_002852
1101 | 2010_002885
1102 | 2010_002889
1103 | 2010_002904
1104 | 2010_002908
1105 | 2010_002916
1106 | 2010_002974
1107 | 2010_002977
1108 | 2010_003005
1109 | 2010_003021
1110 | 2010_003030
1111 | 2010_003038
1112 | 2010_003046
1113 | 2010_003052
1114 | 2010_003089
1115 | 2010_003110
1116 | 2010_003118
1117 | 2010_003171
1118 | 2010_003217
1119 | 2010_003221
1120 | 2010_003228
1121 | 2010_003243
1122 | 2010_003271
1123 | 2010_003295
1124 | 2010_003306
1125 | 2010_003324
1126 | 2010_003363
1127 | 2010_003382
1128 | 2010_003388
1129 | 2010_003389
1130 | 2010_003392
1131 | 2010_003430
1132 | 2010_003442
1133 | 2010_003459
1134 | 2010_003485
1135 | 2010_003486
1136 | 2010_003500
1137 | 2010_003523
1138 | 2010_003542
1139 | 2010_003552
1140 | 2010_003570
1141 | 2010_003572
1142 | 2010_003586
1143 | 2010_003615
1144 | 2010_003623
1145 | 2010_003657
1146 | 2010_003666
1147 | 2010_003705
1148 | 2010_003710
1149 | 2010_003720
1150 | 2010_003733
1151 | 2010_003750
1152 | 2010_003767
1153 | 2010_003802
1154 | 2010_003809
1155 | 2010_003830
1156 | 2010_003832
1157 | 2010_003836
1158 | 2010_003838
1159 | 2010_003850
1160 | 2010_003867
1161 | 2010_003882
1162 | 2010_003909
1163 | 2010_003922
1164 | 2010_003923
1165 | 2010_003978
1166 | 2010_003989
1167 | 2010_003990
1168 | 2010_004000
1169 | 2010_004003
1170 | 2010_004068
1171 | 2010_004076
1172 | 2010_004117
1173 | 2010_004136
1174 | 2010_004142
1175 | 2010_004195
1176 | 2010_004200
1177 | 2010_004202
1178 | 2010_004232
1179 | 2010_004261
1180 | 2010_004266
1181 | 2010_004273
1182 | 2010_004305
1183 | 2010_004403
1184 | 2010_004433
1185 | 2010_004434
1186 | 2010_004435
1187 | 2010_004438
1188 | 2010_004442
1189 | 2010_004473
1190 | 2010_004482
1191 | 2010_004487
1192 | 2010_004489
1193 | 2010_004512
1194 | 2010_004525
1195 | 2010_004527
1196 | 2010_004532
1197 | 2010_004566
1198 | 2010_004568
1199 | 2010_004579
1200 | 2010_004611
1201 | 2010_004641
1202 | 2010_004688
1203 | 2010_004699
1204 | 2010_004702
1205 | 2010_004716
1206 | 2010_004754
1207 | 2010_004767
1208 | 2010_004776
1209 | 2010_004811
1210 | 2010_004837
1211 | 2010_004839
1212 | 2010_004845
1213 | 2010_004860
1214 | 2010_004867
1215 | 2010_004881
1216 | 2010_004939
1217 | 2010_005001
1218 | 2010_005047
1219 | 2010_005051
1220 | 2010_005091
1221 | 2010_005095
1222 | 2010_005125
1223 | 2010_005140
1224 | 2010_005177
1225 | 2010_005178
1226 | 2010_005194
1227 | 2010_005197
1228 | 2010_005200
1229 | 2010_005205
1230 | 2010_005212
1231 | 2010_005248
1232 | 2010_005294
1233 | 2010_005298
1234 | 2010_005313
1235 | 2010_005324
1236 | 2010_005328
1237 | 2010_005329
1238 | 2010_005380
1239 | 2010_005404
1240 | 2010_005407
1241 | 2010_005411
1242 | 2010_005423
1243 | 2010_005499
1244 | 2010_005509
1245 | 2010_005510
1246 | 2010_005544
1247 | 2010_005549
1248 | 2010_005590
1249 | 2010_005639
1250 | 2010_005699
1251 | 2010_005704
1252 | 2010_005707
1253 | 2010_005711
1254 | 2010_005726
1255 | 2010_005741
1256 | 2010_005765
1257 | 2010_005790
1258 | 2010_005792
1259 | 2010_005797
1260 | 2010_005812
1261 | 2010_005850
1262 | 2010_005861
1263 | 2010_005869
1264 | 2010_005908
1265 | 2010_005915
1266 | 2010_005946
1267 | 2010_005965
1268 | 2010_006044
1269 | 2010_006047
1270 | 2010_006052
1271 | 2010_006081
1272 | 2011_000001
1273 | 2011_000013
1274 | 2011_000014
1275 | 2011_000020
1276 | 2011_000032
1277 | 2011_000042
1278 | 2011_000063
1279 | 2011_000115
1280 | 2011_000120
1281 | 2011_000240
1282 | 2011_000244
1283 | 2011_000254
1284 | 2011_000261
1285 | 2011_000262
1286 | 2011_000271
1287 | 2011_000274
1288 | 2011_000306
1289 | 2011_000311
1290 | 2011_000316
1291 | 2011_000328
1292 | 2011_000351
1293 | 2011_000352
1294 | 2011_000406
1295 | 2011_000414
1296 | 2011_000448
1297 | 2011_000451
1298 | 2011_000470
1299 | 2011_000473
1300 | 2011_000515
1301 | 2011_000537
1302 | 2011_000576
1303 | 2011_000603
1304 | 2011_000616
1305 | 2011_000636
1306 | 2011_000639
1307 | 2011_000654
1308 | 2011_000660
1309 | 2011_000664
1310 | 2011_000667
1311 | 2011_000670
1312 | 2011_000676
1313 | 2011_000721
1314 | 2011_000723
1315 | 2011_000762
1316 | 2011_000766
1317 | 2011_000786
1318 | 2011_000802
1319 | 2011_000810
1320 | 2011_000821
1321 | 2011_000841
1322 | 2011_000844
1323 | 2011_000846
1324 | 2011_000869
1325 | 2011_000890
1326 | 2011_000915
1327 | 2011_000924
1328 | 2011_000937
1329 | 2011_000939
1330 | 2011_000952
1331 | 2011_000968
1332 | 2011_000974
1333 | 2011_001037
1334 | 2011_001072
1335 | 2011_001085
1336 | 2011_001089
1337 | 2011_001090
1338 | 2011_001099
1339 | 2011_001104
1340 | 2011_001112
1341 | 2011_001120
1342 | 2011_001132
1343 | 2011_001151
1344 | 2011_001194
1345 | 2011_001258
1346 | 2011_001274
1347 | 2011_001314
1348 | 2011_001317
1349 | 2011_001321
1350 | 2011_001379
1351 | 2011_001425
1352 | 2011_001431
1353 | 2011_001443
1354 | 2011_001446
1355 | 2011_001452
1356 | 2011_001454
1357 | 2011_001477
1358 | 2011_001509
1359 | 2011_001512
1360 | 2011_001515
1361 | 2011_001528
1362 | 2011_001554
1363 | 2011_001561
1364 | 2011_001580
1365 | 2011_001587
1366 | 2011_001623
1367 | 2011_001648
1368 | 2011_001651
1369 | 2011_001654
1370 | 2011_001684
1371 | 2011_001696
1372 | 2011_001697
1373 | 2011_001760
1374 | 2011_001761
1375 | 2011_001798
1376 | 2011_001807
1377 | 2011_001851
1378 | 2011_001852
1379 | 2011_001853
1380 | 2011_001888
1381 | 2011_001940
1382 | 2011_002014
1383 | 2011_002028
1384 | 2011_002056
1385 | 2011_002061
1386 | 2011_002068
1387 | 2011_002076
1388 | 2011_002090
1389 | 2011_002095
1390 | 2011_002104
1391 | 2011_002136
1392 | 2011_002138
1393 | 2011_002151
1394 | 2011_002153
1395 | 2011_002155
1396 | 2011_002197
1397 | 2011_002198
1398 | 2011_002243
1399 | 2011_002250
1400 | 2011_002257
1401 | 2011_002262
1402 | 2011_002264
1403 | 2011_002296
1404 | 2011_002314
1405 | 2011_002331
1406 | 2011_002333
1407 | 2011_002411
1408 | 2011_002417
1409 | 2011_002425
1410 | 2011_002437
1411 | 2011_002444
1412 | 2011_002445
1413 | 2011_002449
1414 | 2011_002468
1415 | 2011_002469
1416 | 2011_002473
1417 | 2011_002508
1418 | 2011_002523
1419 | 2011_002534
1420 | 2011_002557
1421 | 2011_002564
1422 | 2011_002572
1423 | 2011_002597
1424 | 2011_002622
1425 | 2011_002632
1426 | 2011_002635
1427 | 2011_002643
1428 | 2011_002653
1429 | 2011_002667
1430 | 2011_002681
1431 | 2011_002707
1432 | 2011_002736
1433 | 2011_002759
1434 | 2011_002783
1435 | 2011_002792
1436 | 2011_002799
1437 | 2011_002824
1438 | 2011_002835
1439 | 2011_002866
1440 | 2011_002876
1441 | 2011_002888
1442 | 2011_002894
1443 | 2011_002903
1444 | 2011_002905
1445 | 2011_002986
1446 | 2011_003045
1447 | 2011_003064
1448 | 2011_003070
1449 | 2011_003083
1450 | 2011_003093
1451 | 2011_003096
1452 | 2011_003102
1453 | 2011_003156
1454 | 2011_003170
1455 | 2011_003178
1456 | 2011_003231
1457 |
--------------------------------------------------------------------------------
/data/tfrecord/val.txt:
--------------------------------------------------------------------------------
1 | 2007_000033
2 | 2007_000042
3 | 2007_000061
4 | 2007_000123
5 | 2007_000129
6 | 2007_000175
7 | 2007_000187
8 | 2007_000323
9 | 2007_000332
10 | 2007_000346
11 | 2007_000452
12 | 2007_000464
13 | 2007_000491
14 | 2007_000529
15 | 2007_000559
16 | 2007_000572
17 | 2007_000629
18 | 2007_000636
19 | 2007_000661
20 | 2007_000663
21 | 2007_000676
22 | 2007_000727
23 | 2007_000762
24 | 2007_000783
25 | 2007_000799
26 | 2007_000804
27 | 2007_000830
28 | 2007_000837
29 | 2007_000847
30 | 2007_000862
31 | 2007_000925
32 | 2007_000999
33 | 2007_001154
34 | 2007_001175
35 | 2007_001239
36 | 2007_001284
37 | 2007_001288
38 | 2007_001289
39 | 2007_001299
40 | 2007_001311
41 | 2007_001321
42 | 2007_001377
43 | 2007_001408
44 | 2007_001423
45 | 2007_001430
46 | 2007_001457
47 | 2007_001458
48 | 2007_001526
49 | 2007_001568
50 | 2007_001585
51 | 2007_001586
52 | 2007_001587
53 | 2007_001594
54 | 2007_001630
55 | 2007_001677
56 | 2007_001678
57 | 2007_001717
58 | 2007_001733
59 | 2007_001761
60 | 2007_001763
61 | 2007_001774
62 | 2007_001884
63 | 2007_001955
64 | 2007_002046
65 | 2007_002094
66 | 2007_002119
67 | 2007_002132
68 | 2007_002260
69 | 2007_002266
70 | 2007_002268
71 | 2007_002284
72 | 2007_002376
73 | 2007_002378
74 | 2007_002387
75 | 2007_002400
76 | 2007_002412
77 | 2007_002426
78 | 2007_002427
79 | 2007_002445
80 | 2007_002470
81 | 2007_002539
82 | 2007_002565
83 | 2007_002597
84 | 2007_002618
85 | 2007_002619
86 | 2007_002624
87 | 2007_002643
88 | 2007_002648
89 | 2007_002719
90 | 2007_002728
91 | 2007_002823
92 | 2007_002824
93 | 2007_002852
94 | 2007_002903
95 | 2007_003011
96 | 2007_003020
97 | 2007_003022
98 | 2007_003051
99 | 2007_003088
100 | 2007_003101
101 | 2007_003106
102 | 2007_003110
103 | 2007_003131
104 | 2007_003134
105 | 2007_003137
106 | 2007_003143
107 | 2007_003169
108 | 2007_003188
109 | 2007_003194
110 | 2007_003195
111 | 2007_003201
112 | 2007_003349
113 | 2007_003367
114 | 2007_003373
115 | 2007_003499
116 | 2007_003503
117 | 2007_003506
118 | 2007_003530
119 | 2007_003571
120 | 2007_003587
121 | 2007_003611
122 | 2007_003621
123 | 2007_003682
124 | 2007_003711
125 | 2007_003714
126 | 2007_003742
127 | 2007_003786
128 | 2007_003841
129 | 2007_003848
130 | 2007_003861
131 | 2007_003872
132 | 2007_003917
133 | 2007_003957
134 | 2007_003991
135 | 2007_004033
136 | 2007_004052
137 | 2007_004112
138 | 2007_004121
139 | 2007_004143
140 | 2007_004189
141 | 2007_004190
142 | 2007_004193
143 | 2007_004241
144 | 2007_004275
145 | 2007_004281
146 | 2007_004380
147 | 2007_004392
148 | 2007_004405
149 | 2007_004468
150 | 2007_004483
151 | 2007_004510
152 | 2007_004538
153 | 2007_004558
154 | 2007_004644
155 | 2007_004649
156 | 2007_004712
157 | 2007_004722
158 | 2007_004856
159 | 2007_004866
160 | 2007_004902
161 | 2007_004969
162 | 2007_005058
163 | 2007_005074
164 | 2007_005107
165 | 2007_005114
166 | 2007_005149
167 | 2007_005173
168 | 2007_005281
169 | 2007_005294
170 | 2007_005296
171 | 2007_005304
172 | 2007_005331
173 | 2007_005354
174 | 2007_005358
175 | 2007_005428
176 | 2007_005460
177 | 2007_005469
178 | 2007_005509
179 | 2007_005547
180 | 2007_005600
181 | 2007_005608
182 | 2007_005626
183 | 2007_005689
184 | 2007_005696
185 | 2007_005705
186 | 2007_005759
187 | 2007_005803
188 | 2007_005813
189 | 2007_005828
190 | 2007_005844
191 | 2007_005845
192 | 2007_005857
193 | 2007_005911
194 | 2007_005915
195 | 2007_005978
196 | 2007_006028
197 | 2007_006035
198 | 2007_006046
199 | 2007_006076
200 | 2007_006086
201 | 2007_006117
202 | 2007_006171
203 | 2007_006241
204 | 2007_006260
205 | 2007_006277
206 | 2007_006348
207 | 2007_006364
208 | 2007_006373
209 | 2007_006444
210 | 2007_006449
211 | 2007_006549
212 | 2007_006553
213 | 2007_006560
214 | 2007_006647
215 | 2007_006678
216 | 2007_006680
217 | 2007_006698
218 | 2007_006761
219 | 2007_006802
220 | 2007_006837
221 | 2007_006841
222 | 2007_006864
223 | 2007_006866
224 | 2007_006946
225 | 2007_007007
226 | 2007_007084
227 | 2007_007109
228 | 2007_007130
229 | 2007_007165
230 | 2007_007168
231 | 2007_007195
232 | 2007_007196
233 | 2007_007203
234 | 2007_007211
235 | 2007_007235
236 | 2007_007341
237 | 2007_007414
238 | 2007_007417
239 | 2007_007470
240 | 2007_007477
241 | 2007_007493
242 | 2007_007498
243 | 2007_007524
244 | 2007_007534
245 | 2007_007624
246 | 2007_007651
247 | 2007_007688
248 | 2007_007748
249 | 2007_007795
250 | 2007_007810
251 | 2007_007815
252 | 2007_007818
253 | 2007_007836
254 | 2007_007849
255 | 2007_007881
256 | 2007_007996
257 | 2007_008051
258 | 2007_008084
259 | 2007_008106
260 | 2007_008110
261 | 2007_008204
262 | 2007_008222
263 | 2007_008256
264 | 2007_008260
265 | 2007_008339
266 | 2007_008374
267 | 2007_008415
268 | 2007_008430
269 | 2007_008543
270 | 2007_008547
271 | 2007_008596
272 | 2007_008645
273 | 2007_008670
274 | 2007_008708
275 | 2007_008722
276 | 2007_008747
277 | 2007_008802
278 | 2007_008815
279 | 2007_008897
280 | 2007_008944
281 | 2007_008964
282 | 2007_008973
283 | 2007_008980
284 | 2007_009015
285 | 2007_009068
286 | 2007_009084
287 | 2007_009088
288 | 2007_009096
289 | 2007_009221
290 | 2007_009245
291 | 2007_009251
292 | 2007_009252
293 | 2007_009258
294 | 2007_009320
295 | 2007_009323
296 | 2007_009331
297 | 2007_009346
298 | 2007_009392
299 | 2007_009413
300 | 2007_009419
301 | 2007_009446
302 | 2007_009458
303 | 2007_009521
304 | 2007_009562
305 | 2007_009592
306 | 2007_009654
307 | 2007_009655
308 | 2007_009684
309 | 2007_009687
310 | 2007_009691
311 | 2007_009706
312 | 2007_009750
313 | 2007_009756
314 | 2007_009764
315 | 2007_009794
316 | 2007_009817
317 | 2007_009841
318 | 2007_009897
319 | 2007_009911
320 | 2007_009923
321 | 2007_009938
322 | 2008_000009
323 | 2008_000016
324 | 2008_000073
325 | 2008_000075
326 | 2008_000080
327 | 2008_000107
328 | 2008_000120
329 | 2008_000123
330 | 2008_000149
331 | 2008_000182
332 | 2008_000213
333 | 2008_000215
334 | 2008_000223
335 | 2008_000233
336 | 2008_000234
337 | 2008_000239
338 | 2008_000254
339 | 2008_000270
340 | 2008_000271
341 | 2008_000345
342 | 2008_000359
343 | 2008_000391
344 | 2008_000401
345 | 2008_000464
346 | 2008_000469
347 | 2008_000474
348 | 2008_000501
349 | 2008_000510
350 | 2008_000533
351 | 2008_000573
352 | 2008_000589
353 | 2008_000602
354 | 2008_000630
355 | 2008_000657
356 | 2008_000661
357 | 2008_000662
358 | 2008_000666
359 | 2008_000673
360 | 2008_000700
361 | 2008_000725
362 | 2008_000731
363 | 2008_000763
364 | 2008_000765
365 | 2008_000782
366 | 2008_000795
367 | 2008_000811
368 | 2008_000848
369 | 2008_000853
370 | 2008_000863
371 | 2008_000911
372 | 2008_000919
373 | 2008_000943
374 | 2008_000992
375 | 2008_001013
376 | 2008_001028
377 | 2008_001040
378 | 2008_001070
379 | 2008_001074
380 | 2008_001076
381 | 2008_001078
382 | 2008_001135
383 | 2008_001150
384 | 2008_001170
385 | 2008_001231
386 | 2008_001249
387 | 2008_001260
388 | 2008_001283
389 | 2008_001308
390 | 2008_001379
391 | 2008_001404
392 | 2008_001433
393 | 2008_001439
394 | 2008_001478
395 | 2008_001491
396 | 2008_001504
397 | 2008_001513
398 | 2008_001514
399 | 2008_001531
400 | 2008_001546
401 | 2008_001547
402 | 2008_001580
403 | 2008_001629
404 | 2008_001640
405 | 2008_001682
406 | 2008_001688
407 | 2008_001715
408 | 2008_001821
409 | 2008_001874
410 | 2008_001885
411 | 2008_001895
412 | 2008_001966
413 | 2008_001971
414 | 2008_001992
415 | 2008_002043
416 | 2008_002152
417 | 2008_002205
418 | 2008_002212
419 | 2008_002239
420 | 2008_002240
421 | 2008_002241
422 | 2008_002269
423 | 2008_002273
424 | 2008_002358
425 | 2008_002379
426 | 2008_002383
427 | 2008_002429
428 | 2008_002464
429 | 2008_002467
430 | 2008_002492
431 | 2008_002495
432 | 2008_002504
433 | 2008_002521
434 | 2008_002536
435 | 2008_002588
436 | 2008_002623
437 | 2008_002680
438 | 2008_002681
439 | 2008_002775
440 | 2008_002778
441 | 2008_002835
442 | 2008_002859
443 | 2008_002864
444 | 2008_002900
445 | 2008_002904
446 | 2008_002929
447 | 2008_002936
448 | 2008_002942
449 | 2008_002958
450 | 2008_003003
451 | 2008_003026
452 | 2008_003034
453 | 2008_003076
454 | 2008_003105
455 | 2008_003108
456 | 2008_003110
457 | 2008_003135
458 | 2008_003141
459 | 2008_003155
460 | 2008_003210
461 | 2008_003238
462 | 2008_003270
463 | 2008_003330
464 | 2008_003333
465 | 2008_003369
466 | 2008_003379
467 | 2008_003451
468 | 2008_003461
469 | 2008_003477
470 | 2008_003492
471 | 2008_003499
472 | 2008_003511
473 | 2008_003546
474 | 2008_003576
475 | 2008_003577
476 | 2008_003676
477 | 2008_003709
478 | 2008_003733
479 | 2008_003777
480 | 2008_003782
481 | 2008_003821
482 | 2008_003846
483 | 2008_003856
484 | 2008_003858
485 | 2008_003874
486 | 2008_003876
487 | 2008_003885
488 | 2008_003886
489 | 2008_003926
490 | 2008_003976
491 | 2008_004069
492 | 2008_004101
493 | 2008_004140
494 | 2008_004172
495 | 2008_004175
496 | 2008_004212
497 | 2008_004279
498 | 2008_004339
499 | 2008_004345
500 | 2008_004363
501 | 2008_004367
502 | 2008_004396
503 | 2008_004399
504 | 2008_004453
505 | 2008_004477
506 | 2008_004552
507 | 2008_004562
508 | 2008_004575
509 | 2008_004610
510 | 2008_004612
511 | 2008_004621
512 | 2008_004624
513 | 2008_004654
514 | 2008_004659
515 | 2008_004687
516 | 2008_004701
517 | 2008_004704
518 | 2008_004705
519 | 2008_004754
520 | 2008_004758
521 | 2008_004854
522 | 2008_004910
523 | 2008_004995
524 | 2008_005049
525 | 2008_005089
526 | 2008_005097
527 | 2008_005105
528 | 2008_005145
529 | 2008_005197
530 | 2008_005217
531 | 2008_005242
532 | 2008_005245
533 | 2008_005254
534 | 2008_005262
535 | 2008_005338
536 | 2008_005398
537 | 2008_005399
538 | 2008_005422
539 | 2008_005439
540 | 2008_005445
541 | 2008_005525
542 | 2008_005544
543 | 2008_005628
544 | 2008_005633
545 | 2008_005637
546 | 2008_005642
547 | 2008_005676
548 | 2008_005680
549 | 2008_005691
550 | 2008_005727
551 | 2008_005738
552 | 2008_005812
553 | 2008_005904
554 | 2008_005915
555 | 2008_006008
556 | 2008_006036
557 | 2008_006055
558 | 2008_006063
559 | 2008_006108
560 | 2008_006130
561 | 2008_006143
562 | 2008_006159
563 | 2008_006216
564 | 2008_006219
565 | 2008_006229
566 | 2008_006254
567 | 2008_006275
568 | 2008_006325
569 | 2008_006327
570 | 2008_006341
571 | 2008_006408
572 | 2008_006480
573 | 2008_006523
574 | 2008_006526
575 | 2008_006528
576 | 2008_006553
577 | 2008_006554
578 | 2008_006703
579 | 2008_006722
580 | 2008_006752
581 | 2008_006784
582 | 2008_006835
583 | 2008_006874
584 | 2008_006981
585 | 2008_006986
586 | 2008_007025
587 | 2008_007031
588 | 2008_007048
589 | 2008_007120
590 | 2008_007123
591 | 2008_007143
592 | 2008_007194
593 | 2008_007219
594 | 2008_007273
595 | 2008_007350
596 | 2008_007378
597 | 2008_007392
598 | 2008_007402
599 | 2008_007497
600 | 2008_007498
601 | 2008_007507
602 | 2008_007513
603 | 2008_007527
604 | 2008_007548
605 | 2008_007596
606 | 2008_007677
607 | 2008_007737
608 | 2008_007797
609 | 2008_007804
610 | 2008_007811
611 | 2008_007814
612 | 2008_007828
613 | 2008_007836
614 | 2008_007945
615 | 2008_007994
616 | 2008_008051
617 | 2008_008103
618 | 2008_008127
619 | 2008_008221
620 | 2008_008252
621 | 2008_008268
622 | 2008_008296
623 | 2008_008301
624 | 2008_008335
625 | 2008_008362
626 | 2008_008392
627 | 2008_008393
628 | 2008_008421
629 | 2008_008434
630 | 2008_008469
631 | 2008_008629
632 | 2008_008682
633 | 2008_008711
634 | 2008_008746
635 | 2009_000012
636 | 2009_000013
637 | 2009_000022
638 | 2009_000032
639 | 2009_000037
640 | 2009_000039
641 | 2009_000074
642 | 2009_000080
643 | 2009_000087
644 | 2009_000096
645 | 2009_000121
646 | 2009_000136
647 | 2009_000149
648 | 2009_000156
649 | 2009_000201
650 | 2009_000205
651 | 2009_000219
652 | 2009_000242
653 | 2009_000309
654 | 2009_000318
655 | 2009_000335
656 | 2009_000351
657 | 2009_000354
658 | 2009_000387
659 | 2009_000391
660 | 2009_000412
661 | 2009_000418
662 | 2009_000421
663 | 2009_000426
664 | 2009_000440
665 | 2009_000446
666 | 2009_000455
667 | 2009_000457
668 | 2009_000469
669 | 2009_000487
670 | 2009_000488
671 | 2009_000523
672 | 2009_000573
673 | 2009_000619
674 | 2009_000628
675 | 2009_000641
676 | 2009_000664
677 | 2009_000675
678 | 2009_000704
679 | 2009_000705
680 | 2009_000712
681 | 2009_000716
682 | 2009_000723
683 | 2009_000727
684 | 2009_000730
685 | 2009_000731
686 | 2009_000732
687 | 2009_000771
688 | 2009_000825
689 | 2009_000828
690 | 2009_000839
691 | 2009_000840
692 | 2009_000845
693 | 2009_000879
694 | 2009_000892
695 | 2009_000919
696 | 2009_000924
697 | 2009_000931
698 | 2009_000935
699 | 2009_000964
700 | 2009_000989
701 | 2009_000991
702 | 2009_000998
703 | 2009_001008
704 | 2009_001082
705 | 2009_001108
706 | 2009_001160
707 | 2009_001215
708 | 2009_001240
709 | 2009_001255
710 | 2009_001278
711 | 2009_001299
712 | 2009_001300
713 | 2009_001314
714 | 2009_001332
715 | 2009_001333
716 | 2009_001363
717 | 2009_001391
718 | 2009_001411
719 | 2009_001433
720 | 2009_001505
721 | 2009_001535
722 | 2009_001536
723 | 2009_001565
724 | 2009_001607
725 | 2009_001644
726 | 2009_001663
727 | 2009_001683
728 | 2009_001684
729 | 2009_001687
730 | 2009_001718
731 | 2009_001731
732 | 2009_001765
733 | 2009_001768
734 | 2009_001775
735 | 2009_001804
736 | 2009_001816
737 | 2009_001818
738 | 2009_001850
739 | 2009_001851
740 | 2009_001854
741 | 2009_001941
742 | 2009_001991
743 | 2009_002012
744 | 2009_002035
745 | 2009_002042
746 | 2009_002082
747 | 2009_002094
748 | 2009_002097
749 | 2009_002122
750 | 2009_002150
751 | 2009_002155
752 | 2009_002164
753 | 2009_002165
754 | 2009_002171
755 | 2009_002185
756 | 2009_002202
757 | 2009_002221
758 | 2009_002238
759 | 2009_002239
760 | 2009_002265
761 | 2009_002268
762 | 2009_002291
763 | 2009_002295
764 | 2009_002317
765 | 2009_002320
766 | 2009_002346
767 | 2009_002366
768 | 2009_002372
769 | 2009_002382
770 | 2009_002390
771 | 2009_002415
772 | 2009_002445
773 | 2009_002487
774 | 2009_002521
775 | 2009_002527
776 | 2009_002535
777 | 2009_002539
778 | 2009_002549
779 | 2009_002562
780 | 2009_002568
781 | 2009_002571
782 | 2009_002573
783 | 2009_002584
784 | 2009_002591
785 | 2009_002594
786 | 2009_002604
787 | 2009_002618
788 | 2009_002635
789 | 2009_002638
790 | 2009_002649
791 | 2009_002651
792 | 2009_002727
793 | 2009_002732
794 | 2009_002749
795 | 2009_002753
796 | 2009_002771
797 | 2009_002808
798 | 2009_002856
799 | 2009_002887
800 | 2009_002888
801 | 2009_002928
802 | 2009_002936
803 | 2009_002975
804 | 2009_002982
805 | 2009_002990
806 | 2009_003003
807 | 2009_003005
808 | 2009_003043
809 | 2009_003059
810 | 2009_003063
811 | 2009_003065
812 | 2009_003071
813 | 2009_003080
814 | 2009_003105
815 | 2009_003123
816 | 2009_003193
817 | 2009_003196
818 | 2009_003217
819 | 2009_003224
820 | 2009_003241
821 | 2009_003269
822 | 2009_003273
823 | 2009_003299
824 | 2009_003304
825 | 2009_003311
826 | 2009_003323
827 | 2009_003343
828 | 2009_003378
829 | 2009_003387
830 | 2009_003406
831 | 2009_003433
832 | 2009_003450
833 | 2009_003466
834 | 2009_003481
835 | 2009_003494
836 | 2009_003498
837 | 2009_003504
838 | 2009_003507
839 | 2009_003517
840 | 2009_003523
841 | 2009_003542
842 | 2009_003549
843 | 2009_003551
844 | 2009_003564
845 | 2009_003569
846 | 2009_003576
847 | 2009_003589
848 | 2009_003607
849 | 2009_003640
850 | 2009_003666
851 | 2009_003696
852 | 2009_003703
853 | 2009_003707
854 | 2009_003756
855 | 2009_003771
856 | 2009_003773
857 | 2009_003804
858 | 2009_003806
859 | 2009_003810
860 | 2009_003849
861 | 2009_003857
862 | 2009_003858
863 | 2009_003895
864 | 2009_003903
865 | 2009_003904
866 | 2009_003928
867 | 2009_003938
868 | 2009_003971
869 | 2009_003991
870 | 2009_004021
871 | 2009_004033
872 | 2009_004043
873 | 2009_004070
874 | 2009_004072
875 | 2009_004084
876 | 2009_004099
877 | 2009_004125
878 | 2009_004140
879 | 2009_004217
880 | 2009_004221
881 | 2009_004247
882 | 2009_004248
883 | 2009_004255
884 | 2009_004298
885 | 2009_004324
886 | 2009_004455
887 | 2009_004494
888 | 2009_004497
889 | 2009_004504
890 | 2009_004507
891 | 2009_004509
892 | 2009_004540
893 | 2009_004568
894 | 2009_004579
895 | 2009_004581
896 | 2009_004590
897 | 2009_004592
898 | 2009_004594
899 | 2009_004635
900 | 2009_004653
901 | 2009_004687
902 | 2009_004721
903 | 2009_004730
904 | 2009_004732
905 | 2009_004738
906 | 2009_004748
907 | 2009_004789
908 | 2009_004799
909 | 2009_004801
910 | 2009_004848
911 | 2009_004859
912 | 2009_004867
913 | 2009_004882
914 | 2009_004886
915 | 2009_004895
916 | 2009_004942
917 | 2009_004969
918 | 2009_004987
919 | 2009_004993
920 | 2009_004994
921 | 2009_005038
922 | 2009_005078
923 | 2009_005087
924 | 2009_005089
925 | 2009_005137
926 | 2009_005148
927 | 2009_005156
928 | 2009_005158
929 | 2009_005189
930 | 2009_005190
931 | 2009_005217
932 | 2009_005219
933 | 2009_005220
934 | 2009_005231
935 | 2009_005260
936 | 2009_005262
937 | 2009_005302
938 | 2010_000003
939 | 2010_000038
940 | 2010_000065
941 | 2010_000083
942 | 2010_000084
943 | 2010_000087
944 | 2010_000110
945 | 2010_000159
946 | 2010_000160
947 | 2010_000163
948 | 2010_000174
949 | 2010_000216
950 | 2010_000238
951 | 2010_000241
952 | 2010_000256
953 | 2010_000272
954 | 2010_000284
955 | 2010_000309
956 | 2010_000318
957 | 2010_000330
958 | 2010_000335
959 | 2010_000342
960 | 2010_000372
961 | 2010_000422
962 | 2010_000426
963 | 2010_000427
964 | 2010_000502
965 | 2010_000530
966 | 2010_000552
967 | 2010_000559
968 | 2010_000572
969 | 2010_000573
970 | 2010_000622
971 | 2010_000628
972 | 2010_000639
973 | 2010_000666
974 | 2010_000679
975 | 2010_000682
976 | 2010_000683
977 | 2010_000724
978 | 2010_000738
979 | 2010_000764
980 | 2010_000788
981 | 2010_000814
982 | 2010_000836
983 | 2010_000874
984 | 2010_000904
985 | 2010_000906
986 | 2010_000907
987 | 2010_000918
988 | 2010_000929
989 | 2010_000941
990 | 2010_000952
991 | 2010_000961
992 | 2010_001000
993 | 2010_001010
994 | 2010_001011
995 | 2010_001016
996 | 2010_001017
997 | 2010_001024
998 | 2010_001036
999 | 2010_001061
1000 | 2010_001069
1001 | 2010_001070
1002 | 2010_001079
1003 | 2010_001104
1004 | 2010_001124
1005 | 2010_001149
1006 | 2010_001151
1007 | 2010_001174
1008 | 2010_001206
1009 | 2010_001246
1010 | 2010_001251
1011 | 2010_001256
1012 | 2010_001264
1013 | 2010_001292
1014 | 2010_001313
1015 | 2010_001327
1016 | 2010_001331
1017 | 2010_001351
1018 | 2010_001367
1019 | 2010_001376
1020 | 2010_001403
1021 | 2010_001448
1022 | 2010_001451
1023 | 2010_001522
1024 | 2010_001534
1025 | 2010_001553
1026 | 2010_001557
1027 | 2010_001563
1028 | 2010_001577
1029 | 2010_001579
1030 | 2010_001646
1031 | 2010_001656
1032 | 2010_001692
1033 | 2010_001699
1034 | 2010_001734
1035 | 2010_001752
1036 | 2010_001767
1037 | 2010_001768
1038 | 2010_001773
1039 | 2010_001820
1040 | 2010_001830
1041 | 2010_001851
1042 | 2010_001908
1043 | 2010_001913
1044 | 2010_001951
1045 | 2010_001956
1046 | 2010_001962
1047 | 2010_001966
1048 | 2010_001995
1049 | 2010_002017
1050 | 2010_002025
1051 | 2010_002030
1052 | 2010_002106
1053 | 2010_002137
1054 | 2010_002142
1055 | 2010_002146
1056 | 2010_002147
1057 | 2010_002150
1058 | 2010_002161
1059 | 2010_002200
1060 | 2010_002228
1061 | 2010_002232
1062 | 2010_002251
1063 | 2010_002271
1064 | 2010_002305
1065 | 2010_002310
1066 | 2010_002336
1067 | 2010_002348
1068 | 2010_002361
1069 | 2010_002390
1070 | 2010_002396
1071 | 2010_002422
1072 | 2010_002450
1073 | 2010_002480
1074 | 2010_002512
1075 | 2010_002531
1076 | 2010_002536
1077 | 2010_002538
1078 | 2010_002546
1079 | 2010_002623
1080 | 2010_002682
1081 | 2010_002691
1082 | 2010_002693
1083 | 2010_002701
1084 | 2010_002763
1085 | 2010_002792
1086 | 2010_002868
1087 | 2010_002900
1088 | 2010_002902
1089 | 2010_002921
1090 | 2010_002929
1091 | 2010_002939
1092 | 2010_002988
1093 | 2010_003014
1094 | 2010_003060
1095 | 2010_003123
1096 | 2010_003127
1097 | 2010_003132
1098 | 2010_003168
1099 | 2010_003183
1100 | 2010_003187
1101 | 2010_003207
1102 | 2010_003231
1103 | 2010_003239
1104 | 2010_003275
1105 | 2010_003276
1106 | 2010_003293
1107 | 2010_003302
1108 | 2010_003325
1109 | 2010_003362
1110 | 2010_003365
1111 | 2010_003381
1112 | 2010_003402
1113 | 2010_003409
1114 | 2010_003418
1115 | 2010_003446
1116 | 2010_003453
1117 | 2010_003468
1118 | 2010_003473
1119 | 2010_003495
1120 | 2010_003506
1121 | 2010_003514
1122 | 2010_003531
1123 | 2010_003532
1124 | 2010_003541
1125 | 2010_003547
1126 | 2010_003597
1127 | 2010_003675
1128 | 2010_003708
1129 | 2010_003716
1130 | 2010_003746
1131 | 2010_003758
1132 | 2010_003764
1133 | 2010_003768
1134 | 2010_003771
1135 | 2010_003772
1136 | 2010_003781
1137 | 2010_003813
1138 | 2010_003820
1139 | 2010_003854
1140 | 2010_003912
1141 | 2010_003915
1142 | 2010_003947
1143 | 2010_003956
1144 | 2010_003971
1145 | 2010_004041
1146 | 2010_004042
1147 | 2010_004056
1148 | 2010_004063
1149 | 2010_004104
1150 | 2010_004120
1151 | 2010_004149
1152 | 2010_004165
1153 | 2010_004208
1154 | 2010_004219
1155 | 2010_004226
1156 | 2010_004314
1157 | 2010_004320
1158 | 2010_004322
1159 | 2010_004337
1160 | 2010_004348
1161 | 2010_004355
1162 | 2010_004369
1163 | 2010_004382
1164 | 2010_004419
1165 | 2010_004432
1166 | 2010_004472
1167 | 2010_004479
1168 | 2010_004519
1169 | 2010_004520
1170 | 2010_004529
1171 | 2010_004543
1172 | 2010_004550
1173 | 2010_004551
1174 | 2010_004556
1175 | 2010_004559
1176 | 2010_004628
1177 | 2010_004635
1178 | 2010_004662
1179 | 2010_004697
1180 | 2010_004757
1181 | 2010_004763
1182 | 2010_004772
1183 | 2010_004783
1184 | 2010_004789
1185 | 2010_004795
1186 | 2010_004815
1187 | 2010_004825
1188 | 2010_004828
1189 | 2010_004856
1190 | 2010_004857
1191 | 2010_004861
1192 | 2010_004941
1193 | 2010_004946
1194 | 2010_004951
1195 | 2010_004980
1196 | 2010_004994
1197 | 2010_005013
1198 | 2010_005021
1199 | 2010_005046
1200 | 2010_005063
1201 | 2010_005108
1202 | 2010_005118
1203 | 2010_005159
1204 | 2010_005160
1205 | 2010_005166
1206 | 2010_005174
1207 | 2010_005180
1208 | 2010_005187
1209 | 2010_005206
1210 | 2010_005245
1211 | 2010_005252
1212 | 2010_005284
1213 | 2010_005305
1214 | 2010_005344
1215 | 2010_005353
1216 | 2010_005366
1217 | 2010_005401
1218 | 2010_005421
1219 | 2010_005428
1220 | 2010_005432
1221 | 2010_005433
1222 | 2010_005496
1223 | 2010_005501
1224 | 2010_005508
1225 | 2010_005531
1226 | 2010_005534
1227 | 2010_005575
1228 | 2010_005582
1229 | 2010_005606
1230 | 2010_005626
1231 | 2010_005644
1232 | 2010_005664
1233 | 2010_005705
1234 | 2010_005706
1235 | 2010_005709
1236 | 2010_005718
1237 | 2010_005719
1238 | 2010_005727
1239 | 2010_005762
1240 | 2010_005788
1241 | 2010_005860
1242 | 2010_005871
1243 | 2010_005877
1244 | 2010_005888
1245 | 2010_005899
1246 | 2010_005922
1247 | 2010_005991
1248 | 2010_005992
1249 | 2010_006026
1250 | 2010_006034
1251 | 2010_006054
1252 | 2010_006070
1253 | 2011_000045
1254 | 2011_000051
1255 | 2011_000054
1256 | 2011_000066
1257 | 2011_000070
1258 | 2011_000112
1259 | 2011_000173
1260 | 2011_000178
1261 | 2011_000185
1262 | 2011_000226
1263 | 2011_000234
1264 | 2011_000238
1265 | 2011_000239
1266 | 2011_000248
1267 | 2011_000283
1268 | 2011_000291
1269 | 2011_000310
1270 | 2011_000312
1271 | 2011_000338
1272 | 2011_000396
1273 | 2011_000412
1274 | 2011_000419
1275 | 2011_000435
1276 | 2011_000436
1277 | 2011_000438
1278 | 2011_000455
1279 | 2011_000456
1280 | 2011_000479
1281 | 2011_000481
1282 | 2011_000482
1283 | 2011_000503
1284 | 2011_000512
1285 | 2011_000521
1286 | 2011_000526
1287 | 2011_000536
1288 | 2011_000548
1289 | 2011_000566
1290 | 2011_000585
1291 | 2011_000598
1292 | 2011_000607
1293 | 2011_000618
1294 | 2011_000638
1295 | 2011_000658
1296 | 2011_000661
1297 | 2011_000669
1298 | 2011_000747
1299 | 2011_000780
1300 | 2011_000789
1301 | 2011_000807
1302 | 2011_000809
1303 | 2011_000813
1304 | 2011_000830
1305 | 2011_000843
1306 | 2011_000874
1307 | 2011_000888
1308 | 2011_000900
1309 | 2011_000912
1310 | 2011_000953
1311 | 2011_000969
1312 | 2011_001005
1313 | 2011_001014
1314 | 2011_001020
1315 | 2011_001047
1316 | 2011_001060
1317 | 2011_001064
1318 | 2011_001069
1319 | 2011_001071
1320 | 2011_001082
1321 | 2011_001110
1322 | 2011_001114
1323 | 2011_001159
1324 | 2011_001161
1325 | 2011_001190
1326 | 2011_001232
1327 | 2011_001263
1328 | 2011_001276
1329 | 2011_001281
1330 | 2011_001287
1331 | 2011_001292
1332 | 2011_001313
1333 | 2011_001341
1334 | 2011_001346
1335 | 2011_001350
1336 | 2011_001407
1337 | 2011_001416
1338 | 2011_001421
1339 | 2011_001434
1340 | 2011_001447
1341 | 2011_001489
1342 | 2011_001529
1343 | 2011_001530
1344 | 2011_001534
1345 | 2011_001546
1346 | 2011_001567
1347 | 2011_001589
1348 | 2011_001597
1349 | 2011_001601
1350 | 2011_001607
1351 | 2011_001613
1352 | 2011_001614
1353 | 2011_001619
1354 | 2011_001624
1355 | 2011_001642
1356 | 2011_001665
1357 | 2011_001669
1358 | 2011_001674
1359 | 2011_001708
1360 | 2011_001713
1361 | 2011_001714
1362 | 2011_001722
1363 | 2011_001726
1364 | 2011_001745
1365 | 2011_001748
1366 | 2011_001775
1367 | 2011_001782
1368 | 2011_001793
1369 | 2011_001794
1370 | 2011_001812
1371 | 2011_001862
1372 | 2011_001863
1373 | 2011_001868
1374 | 2011_001880
1375 | 2011_001910
1376 | 2011_001984
1377 | 2011_001988
1378 | 2011_002002
1379 | 2011_002040
1380 | 2011_002041
1381 | 2011_002064
1382 | 2011_002075
1383 | 2011_002098
1384 | 2011_002110
1385 | 2011_002121
1386 | 2011_002124
1387 | 2011_002150
1388 | 2011_002156
1389 | 2011_002178
1390 | 2011_002200
1391 | 2011_002223
1392 | 2011_002244
1393 | 2011_002247
1394 | 2011_002279
1395 | 2011_002295
1396 | 2011_002298
1397 | 2011_002308
1398 | 2011_002317
1399 | 2011_002322
1400 | 2011_002327
1401 | 2011_002343
1402 | 2011_002358
1403 | 2011_002371
1404 | 2011_002379
1405 | 2011_002391
1406 | 2011_002498
1407 | 2011_002509
1408 | 2011_002515
1409 | 2011_002532
1410 | 2011_002535
1411 | 2011_002548
1412 | 2011_002575
1413 | 2011_002578
1414 | 2011_002589
1415 | 2011_002592
1416 | 2011_002623
1417 | 2011_002641
1418 | 2011_002644
1419 | 2011_002662
1420 | 2011_002675
1421 | 2011_002685
1422 | 2011_002713
1423 | 2011_002730
1424 | 2011_002754
1425 | 2011_002812
1426 | 2011_002863
1427 | 2011_002879
1428 | 2011_002885
1429 | 2011_002929
1430 | 2011_002951
1431 | 2011_002975
1432 | 2011_002993
1433 | 2011_002997
1434 | 2011_003003
1435 | 2011_003011
1436 | 2011_003019
1437 | 2011_003030
1438 | 2011_003055
1439 | 2011_003085
1440 | 2011_003103
1441 | 2011_003114
1442 | 2011_003145
1443 | 2011_003146
1444 | 2011_003182
1445 | 2011_003197
1446 | 2011_003205
1447 | 2011_003240
1448 | 2011_003256
1449 | 2011_003271
1450 |
--------------------------------------------------------------------------------
/output/1.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/LeslieZhoa/tensorflow-deeplab_v3_plus/6c4a0788198688da5abf4a91656d7a70a023daed/output/1.jpg
--------------------------------------------------------------------------------
/output/2.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/LeslieZhoa/tensorflow-deeplab_v3_plus/6c4a0788198688da5abf4a91656d7a70a023daed/output/2.jpg
--------------------------------------------------------------------------------
/output/2007_000027.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/LeslieZhoa/tensorflow-deeplab_v3_plus/6c4a0788198688da5abf4a91656d7a70a023daed/output/2007_000027.jpg
--------------------------------------------------------------------------------
/output/2007_000121.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/LeslieZhoa/tensorflow-deeplab_v3_plus/6c4a0788198688da5abf4a91656d7a70a023daed/output/2007_000121.jpg
--------------------------------------------------------------------------------
/output/2007_000243.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/LeslieZhoa/tensorflow-deeplab_v3_plus/6c4a0788198688da5abf4a91656d7a70a023daed/output/2007_000243.jpg
--------------------------------------------------------------------------------
/output/2007_000346.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/LeslieZhoa/tensorflow-deeplab_v3_plus/6c4a0788198688da5abf4a91656d7a70a023daed/output/2007_000346.jpg
--------------------------------------------------------------------------------
/output/2007_000364.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/LeslieZhoa/tensorflow-deeplab_v3_plus/6c4a0788198688da5abf4a91656d7a70a023daed/output/2007_000364.jpg
--------------------------------------------------------------------------------
/output/2007_000452.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/LeslieZhoa/tensorflow-deeplab_v3_plus/6c4a0788198688da5abf4a91656d7a70a023daed/output/2007_000452.jpg
--------------------------------------------------------------------------------
/output/2007_000464.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/LeslieZhoa/tensorflow-deeplab_v3_plus/6c4a0788198688da5abf4a91656d7a70a023daed/output/2007_000464.jpg
--------------------------------------------------------------------------------
/output/2007_000529.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/LeslieZhoa/tensorflow-deeplab_v3_plus/6c4a0788198688da5abf4a91656d7a70a023daed/output/2007_000529.jpg
--------------------------------------------------------------------------------
/output/3.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/LeslieZhoa/tensorflow-deeplab_v3_plus/6c4a0788198688da5abf4a91656d7a70a023daed/output/3.jpg
--------------------------------------------------------------------------------
/output/4.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/LeslieZhoa/tensorflow-deeplab_v3_plus/6c4a0788198688da5abf4a91656d7a70a023daed/output/4.jpg
--------------------------------------------------------------------------------
/output/5.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/LeslieZhoa/tensorflow-deeplab_v3_plus/6c4a0788198688da5abf4a91656d7a70a023daed/output/5.jpg
--------------------------------------------------------------------------------
/output/UNET.png:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/LeslieZhoa/tensorflow-deeplab_v3_plus/6c4a0788198688da5abf4a91656d7a70a023daed/output/UNET.png
--------------------------------------------------------------------------------
/output/model1.png:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/LeslieZhoa/tensorflow-deeplab_v3_plus/6c4a0788198688da5abf4a91656d7a70a023daed/output/model1.png
--------------------------------------------------------------------------------
/output/model2.png:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/LeslieZhoa/tensorflow-deeplab_v3_plus/6c4a0788198688da5abf4a91656d7a70a023daed/output/model2.png
--------------------------------------------------------------------------------
/output/test.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/LeslieZhoa/tensorflow-deeplab_v3_plus/6c4a0788198688da5abf4a91656d7a70a023daed/output/test.jpg
--------------------------------------------------------------------------------
/output/u=3476430323,4263663876&fm=11&gp=0.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/LeslieZhoa/tensorflow-deeplab_v3_plus/6c4a0788198688da5abf4a91656d7a70a023daed/output/u=3476430323,4263663876&fm=11&gp=0.jpg
--------------------------------------------------------------------------------
/output/u=3893090294,1830313637&fm=11&gp=0.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/LeslieZhoa/tensorflow-deeplab_v3_plus/6c4a0788198688da5abf4a91656d7a70a023daed/output/u=3893090294,1830313637&fm=11&gp=0.jpg
--------------------------------------------------------------------------------
/output/w475_h331_9a5169d0369e4e1496d1cdfabb1ded85.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/LeslieZhoa/tensorflow-deeplab_v3_plus/6c4a0788198688da5abf4a91656d7a70a023daed/output/w475_h331_9a5169d0369e4e1496d1cdfabb1ded85.jpg
--------------------------------------------------------------------------------
/picture/1.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/LeslieZhoa/tensorflow-deeplab_v3_plus/6c4a0788198688da5abf4a91656d7a70a023daed/picture/1.jpg
--------------------------------------------------------------------------------
/picture/2.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/LeslieZhoa/tensorflow-deeplab_v3_plus/6c4a0788198688da5abf4a91656d7a70a023daed/picture/2.jpg
--------------------------------------------------------------------------------
/picture/2007_000027.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/LeslieZhoa/tensorflow-deeplab_v3_plus/6c4a0788198688da5abf4a91656d7a70a023daed/picture/2007_000027.jpg
--------------------------------------------------------------------------------
/picture/2007_000121.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/LeslieZhoa/tensorflow-deeplab_v3_plus/6c4a0788198688da5abf4a91656d7a70a023daed/picture/2007_000121.jpg
--------------------------------------------------------------------------------
/picture/2007_000243.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/LeslieZhoa/tensorflow-deeplab_v3_plus/6c4a0788198688da5abf4a91656d7a70a023daed/picture/2007_000243.jpg
--------------------------------------------------------------------------------
/picture/2007_000346.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/LeslieZhoa/tensorflow-deeplab_v3_plus/6c4a0788198688da5abf4a91656d7a70a023daed/picture/2007_000346.jpg
--------------------------------------------------------------------------------
/picture/2007_000364.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/LeslieZhoa/tensorflow-deeplab_v3_plus/6c4a0788198688da5abf4a91656d7a70a023daed/picture/2007_000364.jpg
--------------------------------------------------------------------------------
/picture/2007_000452.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/LeslieZhoa/tensorflow-deeplab_v3_plus/6c4a0788198688da5abf4a91656d7a70a023daed/picture/2007_000452.jpg
--------------------------------------------------------------------------------
/picture/2007_000464.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/LeslieZhoa/tensorflow-deeplab_v3_plus/6c4a0788198688da5abf4a91656d7a70a023daed/picture/2007_000464.jpg
--------------------------------------------------------------------------------
/picture/2007_000529.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/LeslieZhoa/tensorflow-deeplab_v3_plus/6c4a0788198688da5abf4a91656d7a70a023daed/picture/2007_000529.jpg
--------------------------------------------------------------------------------
/picture/3.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/LeslieZhoa/tensorflow-deeplab_v3_plus/6c4a0788198688da5abf4a91656d7a70a023daed/picture/3.jpg
--------------------------------------------------------------------------------
/picture/4.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/LeslieZhoa/tensorflow-deeplab_v3_plus/6c4a0788198688da5abf4a91656d7a70a023daed/picture/4.jpg
--------------------------------------------------------------------------------
/picture/5.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/LeslieZhoa/tensorflow-deeplab_v3_plus/6c4a0788198688da5abf4a91656d7a70a023daed/picture/5.jpg
--------------------------------------------------------------------------------
/picture/test.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/LeslieZhoa/tensorflow-deeplab_v3_plus/6c4a0788198688da5abf4a91656d7a70a023daed/picture/test.jpg
--------------------------------------------------------------------------------
/picture/u=3476430323,4263663876&fm=11&gp=0.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/LeslieZhoa/tensorflow-deeplab_v3_plus/6c4a0788198688da5abf4a91656d7a70a023daed/picture/u=3476430323,4263663876&fm=11&gp=0.jpg
--------------------------------------------------------------------------------
/picture/u=3893090294,1830313637&fm=11&gp=0.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/LeslieZhoa/tensorflow-deeplab_v3_plus/6c4a0788198688da5abf4a91656d7a70a023daed/picture/u=3893090294,1830313637&fm=11&gp=0.jpg
--------------------------------------------------------------------------------
/picture/w475_h331_9a5169d0369e4e1496d1cdfabb1ded85.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/LeslieZhoa/tensorflow-deeplab_v3_plus/6c4a0788198688da5abf4a91656d7a70a023daed/picture/w475_h331_9a5169d0369e4e1496d1cdfabb1ded85.jpg
--------------------------------------------------------------------------------
/test.py:
--------------------------------------------------------------------------------
1 |
2 | # coding: utf-8
3 |
4 | # In[1]:
5 |
6 |
7 | import os
8 | import tensorflow as tf
9 | from utils import config as FLAGS
10 | from utils import deeplab_model,preprocessing
11 | import numpy as np
12 | import cv2
13 |
14 |
15 | # In[2]:
16 |
17 |
18 | def main():
19 | image=tf.placeholder(tf.float32,[None,None,3])
20 | inputs=preprocessing.mean_image_subtraction(image)
21 | inputs=tf.expand_dims(inputs,axis=0)
22 | model=deeplab_model.model_generator(FLAGS.num_classes,
23 | FLAGS.output_stride,
24 | FLAGS.base_architecture,
25 | FLAGS.pre_trained_model,
26 | None,)
27 | logits=model(inputs,False)
28 |
29 | #预测类别shape[batch,h,w,1]
30 | pred_classes=tf.expand_dims(tf.argmax(logits,axis=3,output_type=tf.int32),axis=3)
31 | #图片上色形式shape[batch,h,w,3]
32 | pred_decoded_labels=tf.py_func(preprocessing.decode_labels,
33 | [pred_classes,1,FLAGS.num_classes],
34 | tf.uint8)
35 | pred_decoded_labels=tf.squeeze(pred_decoded_labels)
36 | saver=tf.train.Saver()
37 | sess=tf.Session()
38 | model_file=tf.train.latest_checkpoint(FLAGS.model_dir)
39 | saver.restore(sess,model_file)
40 | if FLAGS.test_mode=='1':
41 | for filename in os.listdir(FLAGS.pictue):
42 | x=cv2.imread(FLAGS.pictue+filename)
43 | x1=cv2.cvtColor(x,cv2.COLOR_BGR2RGB)
44 | out=sess.run(pred_decoded_labels,feed_dict={image:x1})
45 | out=cv2.cvtColor(out,cv2.COLOR_RGB2BGR)
46 | out=np.concatenate([x, out], axis=1)
47 | cv2.imshow('im',out)
48 | k = cv2.waitKey(0) & 0xFF
49 | if k == 27:
50 | cv2.imwrite(FLAGS.output + filename,out)
51 | cv2.destroyAllWindows()
52 |
53 | if FLAGS.test_mode=='2':
54 | cap=cv2.VideoCapture(0)
55 | fourcc = cv2.VideoWriter_fourcc(*'XVID')
56 | out = cv2.VideoWriter(FLAGS.output+'out.mp4' ,fourcc,10,(1280,480))
57 | while True:
58 | ret,frame = cap.read()
59 | if ret == True:
60 | frame1=cv2.cvtColor(frame,cv2.COLOR_BGR2RGB)
61 | result=sess.run(pred_decoded_labels,feed_dict={image:frame1})
62 | result=cv2.cvtColor(result,cv2.COLOR_RGB2BGR)
63 | result1=np.concatenate([frame, result], axis=1)
64 | a = out.write(result1)
65 | cv2.imshow("result", result1)
66 | if cv2.waitKey(1) & 0xFF == ord('q'):
67 | break
68 | else:
69 | break
70 | cap.release()
71 | out.release()
72 | cv2.destroyAllWindows()
73 | sess.close()
74 |
75 |
76 |
77 | # In[3]:
78 |
79 |
80 | if __name__=='__main__':
81 | main()
82 |
83 |
--------------------------------------------------------------------------------
/tfrecord.py:
--------------------------------------------------------------------------------
1 |
2 | # coding: utf-8
3 |
4 | # In[1]:
5 |
6 |
7 | import io
8 | import os
9 | from PIL import Image
10 | import tensorflow as tf
11 | from utils import config as Config
12 | from utils import dataset_util
13 |
14 |
15 | # In[2]:
16 |
17 |
18 | def main():
19 | '''生成tfrecords主程序
20 | '''
21 | if not os.path.exists(Config.tfrecord_path):
22 | os.makedirs(Config.tfrecord_path)
23 | #相当于print
24 | tf.logging.info('读取数据')
25 |
26 | image_dir=os.path.join(Config.data_dir,Config.image_data_dir)
27 | label_dir=os.path.join(Config.data_dir,Config.label_data_dir)
28 |
29 | if not os.path.isdir(label_dir):
30 | raise ValueError('数据缺少，去下载')
31 | #获取训练和验证图片的index
32 | train_examples=dataset_util.read_examples_list(Config.train_data_list)
33 | val_examples=dataset_util.read_examples_list(Config.val_data_list)
34 |
35 | #训练验证tfrecord存储地址
36 | train_output_path=os.path.join(Config.tfrecord_path,'train.record')
37 | val_output_path=os.path.join(Config.tfrecord_path,'val.record')
38 |
39 | #生成tfrecord
40 | create_record(train_output_path,image_dir,label_dir,train_examples)
41 | create_record(val_output_path,image_dir,label_dir,val_examples)
42 |
43 |
44 | # In[3]:
45 |
46 |
47 | def create_record(output_filename,image_dir,label_dir,examples):
48 | '''将图片生成tfrecord
49 | 参数：
50 | output_filename:输出地址
51 | image_dir:图片地址
52 | label_dir:label地址
53 | examples：图片的index名字
54 | '''
55 | writer=tf.python_io.TFRecordWriter(output_filename)
56 | for idx,example in enumerate(examples):
57 | if idx % 500 ==0:
58 | #将生成第几张图片信息输出
59 | tf.logging.info('On image %d of %d',idx,len(examples))
60 | image_path=os.path.join(image_dir,example+'.jpg')
61 | label_path=os.path.join(label_dir,example+'.png')
62 |
63 | if not os.path.exists(image_path):
64 | tf.logging.warning('没有该图片: ',image_path)
65 | continue
66 | elif not os.path.exists(label_path):
67 | tf.logging.warning('没找着label文件： ',label_path)
68 | continue
69 | try:
70 | #转换格式
71 |
72 | tf_example=dict_to_tf_example(image_path,label_path)
73 |
74 | writer.write(tf_example.SerializeToString())
75 | except ValueError:
76 | tf.logging.warning('无效的example： %s, 忽略',example)
77 | writer.close()
78 |
79 |
80 | # In[4]:
81 |
82 |
83 | def dict_to_tf_example(image_path,label_path):
84 | '''格式转换成tfrecord
85 | 参数：
86 | image_path:输入图片地址
87 | label_path:输出label地址
88 | '''
89 | with tf.gfile.GFile(image_path,'rb') as f:
90 | encoder_jpg=f.read()
91 | encoder_jpg_io=io.BytesIO(encoder_jpg)
92 | image=Image.open(encoder_jpg_io)
93 |
94 | if image.format !='JPEG':
95 | tf.logging.info('输入图片格式错误')
96 | raise ValueError('输入图片格式错误')
97 |
98 | with tf.gfile.GFile(label_path,'rb') as f:
99 | encoder_label=f.read()
100 | encoder_label_io=io.BytesIO(encoder_label)
101 | label=Image.open(encoder_label_io)
102 |
103 | if label.format !='PNG':
104 | tf.logging.info('label图片格式错误')
105 | raise ValueError('label图片格式错误')
106 |
107 | if image.size!=label.size:
108 | tf.logging.info('输入输出没对上')
109 | raise ValueError('输入输出没对上')
110 |
111 | example=tf.train.Example(features=tf.train.Features(feature={
112 | 'image':dataset_util.bytes_feature(encoder_jpg),
113 | 'label':dataset_util.bytes_feature(encoder_label)}))
114 | return example
115 |
116 |
117 |
118 | # In[5]:
119 |
120 |
121 | if __name__=='__main__':
122 | #为将要被记录的的东西（日志）设置开始入口
123 | tf.logging.set_verbosity(tf.logging.INFO)
124 | main()
125 |
126 |
--------------------------------------------------------------------------------
/train.py:
--------------------------------------------------------------------------------
1 |
2 | # coding: utf-8
3 |
4 | # In[ ]:
5 |
6 |
7 | import os
8 | import tensorflow as tf
9 | from utils import deeplab_model
10 | from utils import preprocessing
11 | from tensorflow.python import debug as tf_debug
12 | from utils import config as FLAGS
13 | import shutil
14 | _NUM_CLASSES = 21
15 | _HEIGHT = 513
16 | _WIDTH = 513
17 | _DEPTH = 3
18 | _MIN_SCALE = 0.5
19 | _MAX_SCALE = 2.0
20 | _IGNORE_LABEL = 255
21 |
22 | _POWER = 0.9
23 | _MOMENTUM = 0.9
24 |
25 | _BATCH_NORM_DECAY = 0.9997
26 |
27 | _NUM_IMAGES = {
28 | 'train': 10582,
29 | 'validation': 1449,
30 | }
31 |
32 | # In[ ]:
33 |
34 |
35 | def main():
36 | '''训练主函数'''
37 | #设置保存模型时间,和最多保存数目，和多gpu
38 | session_config = tf.ConfigProto(device_count={'GPU': 0,'GPU':1})
39 |
40 |
41 |
42 | run_config=tf.estimator.RunConfig().replace(session_config=session_config,save_checkpoints_secs=1e2, keep_checkpoint_max = 3)
43 |
44 | model=tf.estimator.Estimator(model_fn=deeplab_model.model_fn,
45 | model_dir=FLAGS.model_dir,
46 | config=run_config,
47 | params={
48 | 'output_stride': FLAGS.output_stride,
49 | 'batch_size': FLAGS.batch_size,
50 | 'base_architecture': FLAGS.base_architecture,
51 | 'pre_trained_model': FLAGS.pre_trained_model,
52 | 'batch_norm_decay': _BATCH_NORM_DECAY,
53 | 'num_classes': _NUM_CLASSES,
54 | 'tensorboard_images_max_outputs': FLAGS.tensorboard_images_max_outputs,
55 | 'weight_decay': FLAGS.weight_decay,
56 | 'learning_rate_policy': FLAGS.learning_rate_policy,
57 | 'num_train': _NUM_IMAGES['train'],
58 | 'initial_learning_rate': FLAGS.initial_learning_rate,
59 | 'max_iter': FLAGS.max_iter,
60 | 'end_learning_rate': FLAGS.end_learning_rate,
61 | 'power': _POWER,
62 | 'momentum': _MOMENTUM,
63 | 'freeze_batch_norm': FLAGS.freeze_batch_norm,
64 | 'initial_global_step': FLAGS.initial_global_step
65 | })
66 | for _ in range(FLAGS.train_epochs//FLAGS.epochs_per_eval):
67 | tensors_to_log={
68 | 'global_step':'global_step',
69 | 'learning_rate': 'learning_rate',
70 | 'cross_entropy': 'cross_entropy',
71 | 'train_px_accuracy': 'train_px_accuracy',
72 | 'train_mean_iou': 'train_mean_iou',
73 |
74 | }
75 | #设置训练次数多少输出预测值
76 | loggig_hook=tf.train.LoggingTensorHook(tensors=tensors_to_log,every_n_iter=10)
77 | train_hooks=[loggig_hook]
78 | eval_hooks=None
79 |
80 | if FLAGS.debug:
81 | debug_hook=tf_debug.LocalCLIDebugHook()
82 | train_hooks.append(debug_hook)
83 | eval_hooks=[debug_hook]
84 | tf.logging.info('开始训练里奥')
85 | model.train(input_fn=lambda:input_fn(True,FLAGS.tfrecord_path,FLAGS.batch_size,FLAGS.epochs_per_eval),
86 | hooks=train_hooks)
87 | tf.logging.info('开始验证集里奥')
88 | eval_results=model.evaluate(
89 | input_fn=lambda : input_fn(False,FLAGS.tfrecord_path,1),
90 | hooks=eval_hooks)
91 | print(eval_results)
92 |
93 |
94 | # In[ ]:
95 |
96 |
97 | def input_fn(is_training,data_dir,batch_size,num_epochs=1):
98 | '''将数据搞成estimator输入格式'''
99 | dataset=tf.data.Dataset.from_tensor_slices(get_filenames(is_training,data_dir))
100 | #相当于map
101 | dataset=dataset.flat_map(tf.data.TFRecordDataset)
102 | if is_training:
103 | #打乱
104 | dataset=dataset.shuffle(buffer_size=_NUM_IMAGES['train'])
105 | dataset=dataset.map(parse_record)
106 | dataset=dataset.map(
107 | lambda image,label: preprocess_image(image,label,is_training))
108 | #和batch结合加速
109 | dataset=dataset.prefetch(batch_size)
110 | dataset=dataset.repeat(num_epochs)
111 | dataset=dataset.batch(batch_size)
112 |
113 | iterator=dataset.make_one_shot_iterator()
114 | images,labels=iterator.get_next()
115 | return images,labels
116 |
117 |
118 | # In[ ]:
119 |
120 |
121 | def get_filenames(is_training,data_dir):
122 | '''获取数据目录'''
123 | if is_training:
124 | return [os.path.join(data_dir,'train.record')]
125 | else:
126 | return [os.path.join(data_dir,'val.record')]
127 |
128 |
129 | # In[ ]:
130 |
131 |
132 | def parse_record(raw_record):
133 | '''解析tfrecord数据'''
134 | key_to_features={
135 | 'image':tf.FixedLenFeature((),tf.string,default_value=''),
136 | 'label':tf.FixedLenFeature((),tf.string,default_value='')
137 | }
138 | parsed=tf.parse_single_example(raw_record,key_to_features)
139 | image=tf.image.decode_image(
140 | tf.reshape(parsed['image'],shape=[]),_DEPTH)
141 | image=tf.to_float(tf.image.convert_image_dtype(image,dtype=tf.uint8))
142 | image.set_shape([None,None,3])
143 |
144 | label=tf.image.decode_image(
145 | tf.reshape(parsed['label'],shape=[]),1)
146 | label=tf.to_int32(tf.image.convert_image_dtype(label,dtype=tf.uint8))
147 | label.set_shape([None,None,1])
148 | return image,label
149 |
150 |
151 | # In[ ]:
152 |
153 |
154 | def preprocess_image(image,label,is_training):
155 | '''数据预处理'''
156 | if is_training:
157 | image,label=preprocessing.random_rescale_image_and_label(
158 | image,label,_MIN_SCALE,_MAX_SCALE)
159 | image,label=preprocessing.random_crop_or_pad_image_and_label(
160 | image,label,_HEIGHT,_WIDTH,_IGNORE_LABEL)
161 | image,label=preprocessing.random_filp_left_right_image_and_label(
162 | image,label)
163 | image.set_shape([_HEIGHT,_WIDTH,3])
164 | label.set_shape([_HEIGHT,_WIDTH,1])
165 | image=preprocessing.mean_image_subtraction(image)
166 | return image,label
167 |
168 |
169 | # In[ ]:
170 |
171 |
172 | if __name__=='__main__':
173 | tf.logging.set_verbosity(tf.logging.INFO)
174 | main()
175 |
176 |
--------------------------------------------------------------------------------
/utils/__init__.py:
--------------------------------------------------------------------------------
1 | ###
2 |
--------------------------------------------------------------------------------
/utils/__pycache__/__init__.cpython-35.pyc:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/LeslieZhoa/tensorflow-deeplab_v3_plus/6c4a0788198688da5abf4a91656d7a70a023daed/utils/__pycache__/__init__.cpython-35.pyc
--------------------------------------------------------------------------------
/utils/__pycache__/__init__.cpython-36.pyc:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/LeslieZhoa/tensorflow-deeplab_v3_plus/6c4a0788198688da5abf4a91656d7a70a023daed/utils/__pycache__/__init__.cpython-36.pyc
--------------------------------------------------------------------------------
/utils/__pycache__/config.cpython-35.pyc:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/LeslieZhoa/tensorflow-deeplab_v3_plus/6c4a0788198688da5abf4a91656d7a70a023daed/utils/__pycache__/config.cpython-35.pyc
--------------------------------------------------------------------------------
/utils/__pycache__/config.cpython-36.pyc:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/LeslieZhoa/tensorflow-deeplab_v3_plus/6c4a0788198688da5abf4a91656d7a70a023daed/utils/__pycache__/config.cpython-36.pyc
--------------------------------------------------------------------------------
/utils/__pycache__/dataset_util.cpython-35.pyc:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/LeslieZhoa/tensorflow-deeplab_v3_plus/6c4a0788198688da5abf4a91656d7a70a023daed/utils/__pycache__/dataset_util.cpython-35.pyc
--------------------------------------------------------------------------------
/utils/__pycache__/deeplab1_model.cpython-35.pyc:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/LeslieZhoa/tensorflow-deeplab_v3_plus/6c4a0788198688da5abf4a91656d7a70a023daed/utils/__pycache__/deeplab1_model.cpython-35.pyc
--------------------------------------------------------------------------------
/utils/__pycache__/deeplab_model.cpython-35.pyc:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/LeslieZhoa/tensorflow-deeplab_v3_plus/6c4a0788198688da5abf4a91656d7a70a023daed/utils/__pycache__/deeplab_model.cpython-35.pyc
--------------------------------------------------------------------------------
/utils/__pycache__/deeplab_model.cpython-36.pyc:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/LeslieZhoa/tensorflow-deeplab_v3_plus/6c4a0788198688da5abf4a91656d7a70a023daed/utils/__pycache__/deeplab_model.cpython-36.pyc
--------------------------------------------------------------------------------
/utils/__pycache__/deeplab_model1.cpython-35.pyc:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/LeslieZhoa/tensorflow-deeplab_v3_plus/6c4a0788198688da5abf4a91656d7a70a023daed/utils/__pycache__/deeplab_model1.cpython-35.pyc
--------------------------------------------------------------------------------
/utils/__pycache__/model.cpython-35.pyc:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/LeslieZhoa/tensorflow-deeplab_v3_plus/6c4a0788198688da5abf4a91656d7a70a023daed/utils/__pycache__/model.cpython-35.pyc
--------------------------------------------------------------------------------
/utils/__pycache__/preprocessing.cpython-35.pyc:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/LeslieZhoa/tensorflow-deeplab_v3_plus/6c4a0788198688da5abf4a91656d7a70a023daed/utils/__pycache__/preprocessing.cpython-35.pyc
--------------------------------------------------------------------------------
/utils/__pycache__/preprocessing.cpython-36.pyc:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/LeslieZhoa/tensorflow-deeplab_v3_plus/6c4a0788198688da5abf4a91656d7a70a023daed/utils/__pycache__/preprocessing.cpython-36.pyc
--------------------------------------------------------------------------------
/utils/config.py:
--------------------------------------------------------------------------------
1 | #类别数
2 | num_classes=21
3 | #数据目录
4 | data_dir='./data/VOCdevkit/VOC2012'
5 | #生成tfrecords放置目录
6 | tfrecord_path='./data/tfrecord/'
7 | #训练图片index
8 | train_data_list='./data/train.txt'
9 | #验证图片index
10 | val_data_list='./data/val.txt'
11 | #图片目录
12 | image_data_dir='JPEGImages'
13 | #label目录，每一个像素点即为所分的类别
14 | label_data_dir='SegmentationClassAug'
15 |
16 | #模型目录
17 | model_dir='./model'
18 | #是否清除模型目录
19 | clean_model_dir='store_false'
20 | #训练epoch
21 | train_epochs=2
22 | #训练期间的验证次数
23 | epochs_per_eval=1
24 |
25 | #tensorboard最大图片展示数
26 | tensorboard_images_max_outputs=6
27 |
28 | #批次设置
29 | batch_size=4
30 | #学习率衰减策略
31 | learning_rate_policy='poly'
32 | #学习率衰减最大次数
33 | max_iter=30000
34 |
35 | #重载的结构
36 | base_architecture='resnet_v2_101'
37 | #预训练模型位置
38 | pre_trained_model='./resnet_v2_101/resnet_v2_101.ckpt'
39 | #模型encoder输入与输出比例
40 | output_stride=16
41 | #是否更新BN参数
42 | freeze_batch_norm='store_true'
43 | #起始学习率
44 | initial_learning_rate=7e-3
45 | #终止学习率
46 | end_learning_rate=1e-6
47 | #global_step初始值
48 | initial_global_step=0
49 | #正则化权重
50 | weight_decay=2e-4
51 |
52 | debug=None
53 |
54 | #测试图片地址
55 | pictue='./picture/'
56 | #测试图片输出地址
57 | output='./output/'
58 | #测试输入，若为1则输入图片，为2输入是摄像头
59 | test_mode='1'
60 |
61 |
--------------------------------------------------------------------------------
/utils/dataset_util.py:
--------------------------------------------------------------------------------
1 |
2 | # coding: utf-8
3 |
4 | # In[1]:
5 |
6 |
7 | import tensorflow as tf
8 |
9 |
10 | # In[2]:
11 |
12 |
13 | #tfrecords转换的各种类型
14 | def int_64_feature(value):
15 | return tf.train.Feature(int_64_feature=tf.train.Int64List(value=[value]))
16 | def int64_list_feature(value):
17 | return tf.train.Feature(int64_list=tf.train.Int64List(value=value))
18 |
19 |
20 | def bytes_feature(value):
21 | return tf.train.Feature(bytes_list=tf.train.BytesList(value=[value]))
22 |
23 |
24 | def bytes_list_feature(value):
25 | return tf.train.Feature(bytes_list=tf.train.BytesList(value=value))
26 |
27 |
28 | def float_list_feature(value):
29 | return tf.train.Feature(float_list=tf.train.FloatList(value=value))
30 |
31 |
32 | # In[3]:
33 |
34 |
35 | def read_examples_list(path):
36 | '''返回所有图片的index'''
37 | with tf.gfile.GFile(path) as f:
38 | lines=f.readlines()
39 | return [line.strip().split(' ')[0] for line in lines]
40 |
41 |
42 |
--------------------------------------------------------------------------------
/utils/deeplab_model.py:
--------------------------------------------------------------------------------
1 |
2 | # coding: utf-8
3 |
4 | # In[4]:
5 |
6 |
7 | import tensorflow as tf
8 | from tensorflow.contrib.slim.nets import resnet_v2
9 | from tensorflow.contrib.framework.python.ops import arg_scope
10 | slim=tf.contrib.slim
11 | from utils import preprocessing
12 | _BATH_NORM_DECAY=0.9997
13 | _WEIGHT_DECAY=5e-4
14 |
15 |
16 | # In[10]:
17 |
18 |
19 | def model_fn(features,labels,mode,params):
20 | '''对于estimator的模型接口
21 | 参数：
22 | features:输入特征
23 | labels:真实label
24 | mode:模型模式
25 | params:模型运行相关参数
26 | 返回值：
27 | 模型接口形式
28 | '''
29 | if isinstance(features,dict):
30 | features=features['feature']
31 | #图像加上均值，以便显示
32 | images=tf.cast(tf.map_fn(preprocessing.mean_image_addition,features),
33 | tf.uint8)
34 | network=model_generator(params['num_classes'],
35 | params['output_stride'],
36 | params['base_architecture'],
37 | params['pre_trained_model'],
38 | params['batch_norm_decay'])
39 | logits=network(features,mode==tf.estimator.ModeKeys.TRAIN)
40 | #预测类别shape[batch,h,w,1]
41 | pred_classes=tf.expand_dims(tf.argmax(logits,axis=3,output_type=tf.int32),axis=3)
42 | #图片上色形式shape[batch,h,w,3]
43 | pred_decoded_labels=tf.py_func(preprocessing.decode_labels,
44 | [pred_classes,params['batch_size'],params['num_classes']],
45 | tf.uint8)
46 |
47 | predictions={
48 | 'classes':pred_classes,
49 | 'probabilities':tf.nn.softmax(logits,name='softmax_tensor'),
50 | 'decoded_labels':pred_decoded_labels
51 | }
52 | if mode==tf.estimator.ModeKeys.PREDICT:
53 | #模式为预测，将decoded_labels删掉
54 | predictions_without_decoded_labels=predictions.copy()
55 | del predictions_without_decoded_labels['decoded_labels']
56 | return tf.estimator.EstimatorSpec(
57 | mode=mode,
58 | predictions=predictions,
59 | export_outputs={
60 | 'preds':tf.estimator.export.PredictOutput(
61 | predictions_without_decoded_labels)
62 | })
63 | #为真实label上色
64 | gt_decoded_labels=tf.py_func(preprocessing.decode_labels,
65 | [labels,params['batch_size'],params['num_classes']],tf.uint8)
66 |
67 | labels=tf.squeeze(labels,axis=3)#[batch,h,w]
68 | logits_by_num_classes=tf.reshape(logits,[-1,params['num_classes']])#[-1,21]
69 | labels_flat=tf.reshape(labels,[-1,])#[-1]
70 | #有类别的像素遮罩
71 | valid_indices=tf.to_int32(labels_flat<=params['num_classes']-1)
72 | #除去不明类别的预测和真实值
73 | valid_logits=tf.dynamic_partition(logits_by_num_classes,valid_indices,num_partitions=2)[1]#[-1,num_classes]
74 | valid_labels=tf.dynamic_partition(labels_flat,valid_indices,num_partitions=2)[1]#[-1]
75 |
76 | pred_flat=tf.reshape(pred_classes,[-1,])#[-1]
77 | valid_preds=tf.dynamic_partition(pred_flat,valid_indices,num_partitions=2)[1]#[-1]
78 | #列代表真实值，行代表预测值的混淆矩阵
79 | confusion_matrix=tf.confusion_matrix(valid_labels,valid_preds,num_classes=params['num_classes'])
80 | predictions['valid_preds']=valid_preds
81 | predictions['valid_labels']=valid_labels
82 | predictions['confusion_maxtrix']=confusion_matrix
83 |
84 | #损失函数为交叉熵
85 | cross_entropy=tf.losses.sparse_softmax_cross_entropy(
86 | logits=valid_logits,labels=valid_labels)
87 |
88 | #记录信息
89 | tf.identity(cross_entropy,name='cross_entropy')
90 | tf.summary.scalar('cross_entropy',cross_entropy)
91 |
92 | #训不训练BN里的数值
93 | if not params['freeze_batch_norm']:
94 | train_var_list=[v for v in tf.trainable_variables()]
95 | else:
96 | train_var_list=[v for v in tf.trainable_variables()
97 | if 'beta' not in v.name and 'gamma' not in v.name]
98 | #加上正则计算总损失
99 | with tf.variable_scope('total_loss'):
100 | loss=cross_entropy+params.get('weight_decay',_WEIGHT_DECAY)*tf.add_n(
101 | [tf.nn.l2_loss(v) for v in train_var_list])
102 |
103 | #加入图片到tensrboard
104 | if mode==tf.estimator.ModeKeys.TRAIN:
105 | tf.summary.image('image',
106 | tf.concat(axis=2,values=[images,gt_decoded_labels,pred_decoded_labels]),
107 | max_outputs=params['tensorboard_images_max_outputs'])
108 | global_step=tf.train.get_or_create_global_step()
109 | #选择学习率衰减模式
110 | if params['learning_rate_policy']=='piecewise':
111 | initial_learning_rate=0.1*params['batch_size']/128
112 | #每一个epoch有几个batch
113 | batches_per_epoch=params['num_train']/params['batch_size']
114 | boundaries=[int(batches_per_epoch*epoch) for epoch in [100,150,200]]
115 | values=[initial_learning_rate*decay for decay in [1,0.1,0.01,0.001]]
116 | learning_rate=tf.train.piecewise_constant(
117 | tf.cast(global_step,tf.int32),boundaries,values)
118 | elif params['learning_rate_policy']=='poly':
119 | learning_rate=tf.train.polynomial_decay(
120 | params['initial_learning_rate'],
121 | tf.cast(global_step,tf.int32)-params['initial_global_step'],
122 | params['max_iter'],params['end_learning_rate'],power=params['power'])
123 | else:
124 | raise ValueError('选择一个学习率模型啊')
125 | tf.identity(learning_rate,name='learning_rate')
126 | tf.summary.scalar('learning_rate',learning_rate)
127 |
128 | tf.identity(global_step,name='global_step')
129 | tf.summary.scalar('global_step',global_step)
130 | optimizer=tf.train.MomentumOptimizer(learning_rate=learning_rate,
131 | momentum=params['momentum'])
132 | #BN需相关更新
133 | update_ops=tf.get_collection(tf.GraphKeys.UPDATE_OPS)
134 | with tf.control_dependencies(update_ops):
135 | train_op=optimizer.minimize(loss,global_step,var_list=train_var_list)
136 | else:
137 | train_op=None
138 |
139 | #准确率和平均iou计算
140 | accuracy=tf.metrics.accuracy(valid_labels,valid_preds)
141 | mean_iou=tf.metrics.mean_iou(valid_labels,valid_preds,params['num_classes'])
142 | metrics={'px_accuracy':accuracy,'mean_iou':mean_iou}
143 |
144 | tf.identity(accuracy[1],name='train_px_accuracy')
145 | tf.summary.scalar('train_px_accuracy',accuracy[1])
146 |
147 | def compute_mean_iou(total_cm,name='mean_iou'):
148 | '''计算平均iou
149 | 参数：
150 | total_cm：混淆矩阵
151 | 返回值：平均iou
152 | '''
153 | #分别计算按行按列总数，shape[num_classes]
154 | sum_over_row=tf.to_float(tf.reduce_sum(total_cm,0))
155 | sum_over_col=tf.to_float(tf.reduce_sum(total_cm,1))
156 | #计算对角线即预测正确总数
157 | cm_diag=tf.to_float(tf.diag_part(total_cm))
158 | #分母，shape[num_classes]代表每一个类别
159 | denominator=sum_over_row+sum_over_col-cm_diag
160 |
161 | #计算多少类别有预测值
162 | num_valid_entries=tf.reduce_sum(tf.cast(
163 | tf.not_equal(denominator,0),dtype=tf.float32))
164 | #避免分母为0
165 | denominator=tf.where(tf.greater(
166 | denominator,0),denominator,
167 | tf.ones_like(denominator))
168 | iou=tf.div(cm_diag,denominator)
169 |
170 | for i in range(params['num_classes']):
171 | tf.identity(iou[i],name='train_iou_class{}'.format(i))
172 | tf.summary.scalar('train_iou_class{}'.format(i),iou[i])
173 | result=tf.where(
174 | tf.greater(num_valid_entries,0),
175 | tf.reduce_sum(iou,name=name)/num_valid_entries,
176 | 0)
177 | return result
178 | train_mean_iou=compute_mean_iou(mean_iou[1])
179 | tf.identity(train_mean_iou,name='train_mean_iou')
180 | tf.summary.scalar('train_mean_iou',train_mean_iou)
181 | return tf.estimator.EstimatorSpec(
182 | mode=mode,
183 | predictions=predictions,
184 | loss=loss,
185 | train_op=train_op,
186 | eval_metric_ops=metrics)
187 |
188 |
189 | # In[1]:
190 |
191 |
192 | def model_generator(num_classes,output_stride,
193 | base_architecture,
194 | pre_trained_model,
195 | batch_norm_decay,
196 | data_format='channels_last'):
197 | '''模型主程序
198 | 参数：
199 | num_classes：类别
200 | output_stride:resnet的步长还和空洞卷积膨胀系数有关，若为16,系数为[6,12,18],为8,系数翻倍
201 | base_architecture:resnet的重载模型
202 | pre_trained_model:预训练模型目录
203 | batch_norm_decay:BN层的系数
204 | data_format:输入图片的格式，RGB通道在最前还是最后
205 | 返回值：
206 | 返回预测值shape[batch,h,w,num_classes]
207 | '''
208 | if data_format is None:
209 | pass
210 | if batch_norm_decay is None:
211 | batch_norm_decay=_BATH_NORM_DECAY
212 | if base_architecture not in ['resnet_v2_50','resnet_v2_101']:
213 | raise ValueError('重载模型没整对')
214 | if base_architecture =='resnet_v2_50':
215 | base_model=resnet_v2.resnet_v2_50
216 | else:
217 | base_model=resnet_v2.resnet_v2_101
218 | #建立模型
219 | def model(inputs,is_training):
220 | #统一输入格式为RGB通道放最后
221 | if data_format=='channels_first':
222 | inputs=tf.transpose(inputs,[0,3,1,2])
223 |
224 | #重载resnet
225 | with slim.arg_scope(resnet_v2.resnet_arg_scope(batch_norm_decay=batch_norm_decay)):
226 | logits,end_points=base_model(inputs,
227 | num_classes=None,
228 | is_training=is_training,
229 | global_pool=False,
230 | output_stride=output_stride)
231 | if is_training:
232 | #重载权重
233 | exclude=[base_architecture+'/logits','global_step']
234 | variables_to_restore=slim.get_variables_to_restore(exclude=exclude)
235 | tf.train.init_from_checkpoint(pre_trained_model,
236 | {v.name.split(':')[0]: v for v in variables_to_restore})
237 | inputs_size=tf.shape(inputs)[1:3]
238 | #取一个resnet网络节点
239 | net=end_points[base_architecture+'/block4']
240 | #resnet节点经过ASPP作为编码输出
241 | encoder_output=atrous_spatial_pyramid_pooling(net,output_stride,batch_norm_decay,is_training)
242 |
243 | #解码将图片恢复原来大小
244 | with tf.variable_scope('decoder'):
245 | with slim.arg_scope(resnet_v2.resnet_arg_scope(batch_norm_decay=batch_norm_decay)):
246 | # with slim.arg_scope([slim.conv2d],
247 | # weights_initializer=slim.xavier_initializer(),
248 |
249 | # normalizer_fn=slim.batch_norm,
250 | # normalizer_params={'is_training': is_training, 'decay': batch_norm_decay}):
251 | with tf.variable_scope('low_level_features'):
252 | #又搞来一个节点
253 | low_level_features=end_points[base_architecture+'/block1/unit_3/bottleneck_v2/conv1']
254 | low_level_features=slim.conv2d(low_level_features,48,[1,1],stride=1,scope='conv_1x1')
255 | low_level_features_size=tf.shape(low_level_features)[1:3]
256 |
257 | with tf.variable_scope('upsampling_logits'):
258 | #上采样成输入大小
259 | net=tf.image.resize_bilinear(encoder_output,low_level_features_size,name='upsample_1')
260 | net=tf.concat([net,low_level_features],axis=3,name='concat')
261 | net=slim.conv2d(net,256,[3,3],stride=1,scope='conv_3x3_1')
262 | net=slim.conv2d(net,256,[3,3],stride=1,scope='conv_3x3_2')
263 | net=slim.conv2d(net,num_classes,[1,1],activation_fn=None,normalizer_fn=None,scope='conv_1x1')
264 | logits=tf.image.resize_bilinear(net,inputs_size,name='upsample_2')
265 | return logits
266 | return model
267 |
268 |
269 | # In[7]:
270 |
271 |
272 | def atrous_spatial_pyramid_pooling(inputs,output_stride,
273 | batch_norm_decay,is_training,depth=256):
274 | '''实现ASPP
275 | 参数：
276 | inputs：输入四维向量
277 | output_stride：决定空洞卷积膨胀率
278 | batch_norm_decay:同上函数
279 | is_training:是否训练
280 | depth:输出通道数
281 | 返回值：
282 | ASPP后的输出
283 | '''
284 | with tf.variable_scope('aspp'):
285 | if output_stride not in [8,16]:
286 | raise ValueError('out_stride整错了')
287 | #膨胀率
288 | atrous_rates=[6,12,18]
289 | if output_stride ==8:
290 | atrous_rates=[2*rate for rate in atrous_rates]
291 | with slim.arg_scope(resnet_v2.resnet_arg_scope(batch_norm_decay=batch_norm_decay)):
292 | with slim.arg_scope([slim.conv2d],
293 | weights_initializer=slim.xavier_initializer(),
294 |
295 | normalizer_fn=slim.batch_norm,
296 | normalizer_params={'is_training': is_training, 'decay': batch_norm_decay}):
297 | inputs_size=tf.shape(inputs)[1:3]
298 | #slim.conv2d默认激活函数为relu,padding=SAME
299 | conv_1x1=slim.conv2d(inputs,depth,[1,1],stride=1,scope='conv_1x1')
300 | #空洞卷积rate不为1
301 | conv_3x3_1=slim.conv2d(inputs,depth,[3,3],stride=1,rate=atrous_rates[0],scope='conv_3x3_1')
302 | conv_3x3_2=slim.conv2d(inputs,depth,[3,3],stride=1,rate=atrous_rates[1],scope='conv_3x3_2')
303 | conv_3x3_3=slim.conv2d(inputs,depth,[3,3],stride=1,rate=atrous_rates[2],scope='conv_3x3_3')
304 | with tf.variable_scope('image_level_features'):
305 | #池化
306 | image_level_features=tf.reduce_mean(inputs,axis=[1,2],keep_dims=True,name='global_average_pooling')
307 | image_level_features=slim.conv2d(image_level_features,depth,[1,1],stride=1,scope='conv_1x1')
308 | #双线性插值
309 | image_level_features=tf.image.resize_bilinear(image_level_features,inputs_size,name='upsample')
310 | net=tf.concat([conv_1x1,conv_3x3_1,conv_3x3_2,conv_3x3_3,image_level_features],axis=3,name='concat')
311 | return net
312 |
313 |
--------------------------------------------------------------------------------
/utils/preprocessing.py:
--------------------------------------------------------------------------------
1 |
2 | # coding: utf-8
3 |
4 | # In[1]:
5 |
6 |
7 | ''' 主要进行相关数据预处理'''
8 | from PIL import Image
9 | import numpy as np
10 | import tensorflow as tf
11 |
12 | #三色通道的平均值
13 | _R_MEAN=123.68
14 | _G_MEAN=116.78
15 | _B_MEAN=103.94
16 |
17 | #主要为各分类上色
18 | label_colors=[(0,0,0),#0=背景
19 | #1=飞机， 2=自行车， 3=鸟， 4=船， 5=瓶子
20 | (128,0,0),(0,128,0),(128,128,0),(0,0,128),(128,0,128),
21 | #6=公交车， 7=小汽车， 8=猫， 9=椅子， 10=牛
22 | (0,128,128),(128,128,128),(64,0,0),(192,0,0),(64,128,0),
23 | #11=晚饭桌， 12=狗， 13=马， 14=摩托车， 15=人
24 | (192,128,0),(64,0,128),(192,0,128),(64,128,128),(192,128,128),
25 | #16=盆栽， 17=羊， 18=沙发， 19=火车， 20=电视或显示屏
26 | (0,64,0),(128,64,0),(0,192,0),(128,192,0),(0,64,128)]
27 |
28 |
29 | # In[3]:
30 |
31 |
32 | def decode_labels(mask,num_image=1,num_classes=21):
33 | '''给图片上色
34 | 参数：
35 | mask：shape是[batch,h,w,1]像素值为每一个像素点的类别
36 | num_image：每次处理图片的长数
37 | num_classes:分类类别数
38 | 返回值：
39 | 返回被上色的分割图像
40 | '''
41 | n,h,w,c=mask.shape
42 | assert (n>=num_image),'num_image %d 不能比批次 %d 大' %(n,num_image)
43 | outputs=np.zeros((num_image,h,w,3),dtype=np.uint8)
44 | for i in range(num_image):
45 | #定义一个长宽为h,w的rgb图像
46 | img=Image.new('RGB',(len(mask[i,0]),len(mask[i])))
47 | pixels=img.load()
48 | for j_,j in enumerate(mask[i,:,:,0]):
49 | for k_,k in enumerate(j):
50 | #如果类别在区间内，给图片上色
51 | if k=max_scale:
124 | raise ValueError('尺度大小搞错了')
125 | shape=tf.shape(image)
126 | height=tf.to_float(shape[0])
127 | width=tf.to_float(shape[1])
128 | #生成随机尺度
129 | scale=tf.random_uniform([],minval=min_scale,maxval=max_scale,dtype=tf.float32)
130 |
131 | new_height=tf.to_int32(height*scale)
132 | new_width=tf.to_int32(width*scale)
133 | #双线性插值
134 | image=tf.image.resize_images(image,[new_height,new_width],
135 | method=tf.image.ResizeMethod.BILINEAR)
136 | #最近邻
137 | label=tf.image.resize_images(label,[new_height,new_width],
138 | method=tf.image.ResizeMethod.NEAREST_NEIGHBOR)
139 | return image,label
140 |
141 |
142 | # In[9]:
143 |
144 |
145 | def random_crop_or_pad_image_and_label(image,label,crop_height,crop_width,ignore_label):
146 | '''随机裁剪填补图像
147 | 参数：
148 | image:输入图像[h,w,c]
149 | label:输出label[h,w,1]
150 | crop_height,crop_width:新图像尺寸
151 | ignore_label:被忽略的类别
152 | 返回值：
153 | 处理后的image，label
154 | '''
155 | #因为0填充所以要把0减去，否则填充之后原来的0就变成1了
156 | label=label-ignore_label
157 | label=tf.to_float(label)
158 | shape=tf.shape(image)
159 | height=shape[0]
160 | width=shape[1]
161 | image_and_label=tf.concat([image,label],axis=2)
162 | image_and_label_pad=tf.image.pad_to_bounding_box(
163 | image_and_label,0,0,
164 | tf.maximum(crop_height,height),
165 | tf.maximum(crop_width,width))
166 | image_and_label_crop=tf.random_crop(
167 | image_and_label_pad,[crop_height,crop_width,4])
168 | image_crop=image_and_label_crop[:,:,:3]
169 | label_crop=image_and_label_crop[:,:,3:]
170 | label_crop+=ignore_label
171 | label_crop=tf.to_int32(label_crop)
172 | return image_crop,label_crop
173 |
174 |
175 | # In[10]:
176 |
177 |
178 | def random_filp_left_right_image_and_label(image,label):
179 | '''随机左右翻转图像
180 | 参数：
181 | image：输入图像[h,w,c]
182 | label:输出label[h,w,1]
183 | 返回值：
184 | 处理后的image，label
185 | '''
186 | uniform_random=tf.random_uniform([],0,1.0)
187 | #对比阈值决定翻转
188 | mirror_cond=tf.less(uniform_random,0.5)
189 | #tf.cond是判断语句依据概率来翻转
190 | image=tf.cond(mirror_cond,lambda: tf.reverse(image,[1]),lambda:image)
191 | label=tf.cond(mirror_cond,lambda:tf.reverse(label,[1]),lambda:label)
192 | return image,label
193 |
194 |
195 |
196 | # In[11]:
197 |
198 |
199 | def eval_input_fn(image_filenames,label_filenames=None,batch_size=1):
200 | '''将图像文件夹处理成模型接收data格式
201 | 参数：
202 | image_filenames:图片目录
203 | label_filenames:测试数据没有label
204 | 把batch_size:测试默认batch为1
205 | 返回值：
206 | data形式的数据包含image和label
207 | '''
208 | #读取文件中的图片
209 | def _parse_function(filename,is_label):
210 | #is_label对于测试数据为None
211 | if not is_label:
212 | image_filename,label_filename=filename,None
213 | else :
214 | image_filename,label_filename=filename
215 | image_string=tf.read_file(image_filename)
216 | image=tf.image.decode_image(image_string)
217 | image=tf.to_float(tf.image.convert_image_dtype(image,dtype=tf.uint8))
218 | image.set_shape([None,None,3])
219 | image=mean_image_subtraction(image)
220 | if not is_label:
221 | return image
222 | else:
223 | label_string = tf.read_file(label_filename)
224 | label = tf.image.decode_image(label_string)
225 | label = tf.to_int32(tf.image.convert_image_dtype(label, dtype=tf.uint8))
226 | label.set_shape([None, None, 1])
227 | return image,label
228 | if label_filenames is None:
229 | input_filenames=image_filenames
230 | else:
231 | input_filenames=(image_filenames,label_filenames)
232 | #生成data格式
233 | dataset=tf.data.Dataset.from_tensor_slices(input_filenames)
234 | if label_filenames is None:
235 | dataset=dataset.map(lambda x: _parse_function(x,False))
236 | else:
237 | dataset=dataset.map(lambda x,y:_parse_function((x,y),True))
238 | dataset=dataset.prefetch(batch_size)#和batch一起用加快处理速度
239 | dataset=dataset.batch(batch_size)
240 | #生成迭代器
241 | iterator=dataset.make_one_shot_iterator()
242 | if label_filenames is None:
243 | images=iterator.get_next()
244 | labels=None
245 | else:
246 | images,labels=iterator.get_next()
247 | return images,labels
248 |
249 |
--------------------------------------------------------------------------------