forked from ZJU-lishuang/yolov5_prune
-
Notifications
You must be signed in to change notification settings - Fork 1
/
shortcut_prune_yolov5s.py
604 lines (525 loc) · 27 KB
/
shortcut_prune_yolov5s.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
from modelsori import *
from utils.utils import *
import numpy as np
from copy import deepcopy
from test import test
from terminaltables import AsciiTable
import time
from utils.prune_utils import *
import argparse
def copy_conv(conv_src,conv_dst):
conv_dst[0] = conv_src.conv
conv_dst[1] = conv_src.bn
conv_dst[2] = conv_src.act
def copy_weight_v4(modelyolov5,model):
focus = list(modelyolov5.model.children())[0]
copy_conv(focus.conv, model.module_list[1])
conv1 = list(modelyolov5.model.children())[1]
copy_conv(conv1, model.module_list[2])
cspnet1 = list(modelyolov5.model.children())[2]
copy_conv(cspnet1.cv2, model.module_list[3])
copy_conv(cspnet1.cv1, model.module_list[5])
copy_conv(cspnet1.m[0].cv1, model.module_list[6])
copy_conv(cspnet1.m[0].cv2, model.module_list[7])
copy_conv(cspnet1.cv3, model.module_list[10])
conv2 = list(modelyolov5.model.children())[3]
copy_conv(conv2, model.module_list[11])
cspnet2 = list(modelyolov5.model.children())[4]
copy_conv(cspnet2.cv2, model.module_list[12])
copy_conv(cspnet2.cv1, model.module_list[14])
copy_conv(cspnet2.m[0].cv1, model.module_list[15])
copy_conv(cspnet2.m[0].cv2, model.module_list[16])
copy_conv(cspnet2.m[1].cv1, model.module_list[18])
copy_conv(cspnet2.m[1].cv2, model.module_list[19])
copy_conv(cspnet2.m[2].cv1, model.module_list[21])
copy_conv(cspnet2.m[2].cv2, model.module_list[22])
copy_conv(cspnet2.cv3, model.module_list[25])
conv3 = list(modelyolov5.model.children())[5]
copy_conv(conv3, model.module_list[26])
cspnet3 = list(modelyolov5.model.children())[6]
copy_conv(cspnet3.cv2, model.module_list[27])
copy_conv(cspnet3.cv1, model.module_list[29])
copy_conv(cspnet3.m[0].cv1, model.module_list[30])
copy_conv(cspnet3.m[0].cv2, model.module_list[31])
copy_conv(cspnet3.m[1].cv1, model.module_list[33])
copy_conv(cspnet3.m[1].cv2, model.module_list[34])
copy_conv(cspnet3.m[2].cv1, model.module_list[36])
copy_conv(cspnet3.m[2].cv2, model.module_list[37])
copy_conv(cspnet3.cv3, model.module_list[40])
conv4 = list(modelyolov5.model.children())[7]
copy_conv(conv4, model.module_list[41])
spp = list(modelyolov5.model.children())[8]
copy_conv(spp.cv1, model.module_list[42])
model.module_list[43] = spp.m[0]
model.module_list[45] = spp.m[1]
model.module_list[47] = spp.m[2]
copy_conv(spp.cv2, model.module_list[49])
cspnet4 = list(modelyolov5.model.children())[9]
copy_conv(cspnet4.cv2, model.module_list[50])
copy_conv(cspnet4.cv1, model.module_list[52])
copy_conv(cspnet4.m[0].cv1, model.module_list[53])
copy_conv(cspnet4.m[0].cv2, model.module_list[54])
copy_conv(cspnet4.cv3, model.module_list[56])
conv5 = list(modelyolov5.model.children())[10]
copy_conv(conv5, model.module_list[57])
upsample1 = list(modelyolov5.model.children())[11]
model.module_list[58] = upsample1
cspnet5 = list(modelyolov5.model.children())[13]
copy_conv(cspnet5.cv2, model.module_list[60])
copy_conv(cspnet5.cv1, model.module_list[62])
copy_conv(cspnet5.m[0].cv1, model.module_list[63])
copy_conv(cspnet5.m[0].cv2, model.module_list[64])
copy_conv(cspnet5.cv3, model.module_list[66])
conv6 = list(modelyolov5.model.children())[14]
copy_conv(conv6, model.module_list[67])
upsample2 = list(modelyolov5.model.children())[15]
model.module_list[68] = upsample2
cspnet6 = list(modelyolov5.model.children())[17]
copy_conv(cspnet6.cv2, model.module_list[70])
copy_conv(cspnet6.cv1, model.module_list[72])
copy_conv(cspnet6.m[0].cv1, model.module_list[73])
copy_conv(cspnet6.m[0].cv2, model.module_list[74])
copy_conv(cspnet6.cv3, model.module_list[76])
conv7 = list(modelyolov5.model.children())[18]
copy_conv(conv7, model.module_list[80])
cspnet7 = list(modelyolov5.model.children())[20]
copy_conv(cspnet7.cv2, model.module_list[82])
copy_conv(cspnet7.cv1, model.module_list[84])
copy_conv(cspnet7.m[0].cv1, model.module_list[85])
copy_conv(cspnet7.m[0].cv2, model.module_list[86])
copy_conv(cspnet7.cv3, model.module_list[88])
conv8 = list(modelyolov5.model.children())[21]
copy_conv(conv8, model.module_list[92])
cspnet8 = list(modelyolov5.model.children())[23]
copy_conv(cspnet8.cv2, model.module_list[94])
copy_conv(cspnet8.cv1, model.module_list[96])
copy_conv(cspnet8.m[0].cv1, model.module_list[97])
copy_conv(cspnet8.m[0].cv2, model.module_list[98])
copy_conv(cspnet8.cv3, model.module_list[100])
detect = list(modelyolov5.model.children())[24]
model.module_list[77][0] = detect.m[0]
model.module_list[89][0] = detect.m[1]
model.module_list[101][0] = detect.m[2]
def copy_weight(modelyolov5,model):
focus = list(modelyolov5.model.children())[0]
model.module_list[1][0] = focus.conv.conv
model.module_list[1][1] = focus.conv.bn
model.module_list[1][2] = focus.conv.act
conv1 = list(modelyolov5.model.children())[1]
model.module_list[2][0] = conv1.conv
model.module_list[2][1] = conv1.bn
model.module_list[2][2] = conv1.act
cspnet1 = list(modelyolov5.model.children())[2]
model.module_list[3][0] = cspnet1.cv2
model.module_list[5][0] = cspnet1.cv1.conv
model.module_list[5][1] = cspnet1.cv1.bn
model.module_list[5][2] = cspnet1.cv1.act
model.module_list[9][0] = cspnet1.cv3
model.module_list[11][0] = cspnet1.bn
model.module_list[11][1] = cspnet1.act
model.module_list[6][0] = cspnet1.m[0].cv1.conv
model.module_list[6][1] = cspnet1.m[0].cv1.bn
model.module_list[6][2] = cspnet1.m[0].cv1.act
model.module_list[7][0] = cspnet1.m[0].cv2.conv
model.module_list[7][1] = cspnet1.m[0].cv2.bn
model.module_list[7][2] = cspnet1.m[0].cv2.act
model.module_list[12][0] = cspnet1.cv4.conv
model.module_list[12][1] = cspnet1.cv4.bn
model.module_list[12][2] = cspnet1.cv4.act
conv2 = list(modelyolov5.model.children())[3]
model.module_list[13][0] = conv2.conv
model.module_list[13][1] = conv2.bn
model.module_list[13][2] = conv2.act
cspnet2 = list(modelyolov5.model.children())[4]
model.module_list[14][0] = cspnet2.cv2
model.module_list[16][0] = cspnet2.cv1.conv
model.module_list[16][1] = cspnet2.cv1.bn
model.module_list[16][2] = cspnet2.cv1.act
model.module_list[26][0] = cspnet2.cv3
model.module_list[28][0] = cspnet2.bn
model.module_list[28][1] = cspnet2.act
model.module_list[29][0] = cspnet2.cv4.conv
model.module_list[29][1] = cspnet2.cv4.bn
model.module_list[29][2] = cspnet2.cv4.act
model.module_list[17][0] = cspnet2.m[0].cv1.conv
model.module_list[17][1] = cspnet2.m[0].cv1.bn
model.module_list[17][2] = cspnet2.m[0].cv1.act
model.module_list[18][0] = cspnet2.m[0].cv2.conv
model.module_list[18][1] = cspnet2.m[0].cv2.bn
model.module_list[18][2] = cspnet2.m[0].cv2.act
model.module_list[20][0] = cspnet2.m[1].cv1.conv
model.module_list[20][1] = cspnet2.m[1].cv1.bn
model.module_list[20][2] = cspnet2.m[1].cv1.act
model.module_list[21][0] = cspnet2.m[1].cv2.conv
model.module_list[21][1] = cspnet2.m[1].cv2.bn
model.module_list[21][2] = cspnet2.m[1].cv2.act
model.module_list[23][0] = cspnet2.m[2].cv1.conv
model.module_list[23][1] = cspnet2.m[2].cv1.bn
model.module_list[23][2] = cspnet2.m[2].cv1.act
model.module_list[24][0] = cspnet2.m[2].cv2.conv
model.module_list[24][1] = cspnet2.m[2].cv2.bn
model.module_list[24][2] = cspnet2.m[2].cv2.act
conv3 = list(modelyolov5.model.children())[5]
model.module_list[30][0] = conv3.conv
model.module_list[30][1] = conv3.bn
model.module_list[30][2] = conv3.act
cspnet3 = list(modelyolov5.model.children())[6]
model.module_list[31][0] = cspnet3.cv2
model.module_list[33][0] = cspnet3.cv1.conv
model.module_list[33][1] = cspnet3.cv1.bn
model.module_list[33][2] = cspnet3.cv1.act
model.module_list[43][0] = cspnet3.cv3
model.module_list[45][0] = cspnet3.bn
model.module_list[45][1] = cspnet3.act
model.module_list[46][0] = cspnet3.cv4.conv
model.module_list[46][1] = cspnet3.cv4.bn
model.module_list[46][2] = cspnet3.cv4.act
model.module_list[34][0] = cspnet3.m[0].cv1.conv
model.module_list[34][1] = cspnet3.m[0].cv1.bn
model.module_list[34][2] = cspnet3.m[0].cv1.act
model.module_list[35][0] = cspnet3.m[0].cv2.conv
model.module_list[35][1] = cspnet3.m[0].cv2.bn
model.module_list[35][2] = cspnet3.m[0].cv2.act
model.module_list[37][0] = cspnet3.m[1].cv1.conv
model.module_list[37][1] = cspnet3.m[1].cv1.bn
model.module_list[37][2] = cspnet3.m[1].cv1.act
model.module_list[38][0] = cspnet3.m[1].cv2.conv
model.module_list[38][1] = cspnet3.m[1].cv2.bn
model.module_list[38][2] = cspnet3.m[1].cv2.act
model.module_list[40][0] = cspnet3.m[2].cv1.conv
model.module_list[40][1] = cspnet3.m[2].cv1.bn
model.module_list[40][2] = cspnet3.m[2].cv1.act
model.module_list[41][0] = cspnet3.m[2].cv2.conv
model.module_list[41][1] = cspnet3.m[2].cv2.bn
model.module_list[41][2] = cspnet3.m[2].cv2.act
conv4 = list(modelyolov5.model.children())[7]
model.module_list[47][0] = conv4.conv
model.module_list[47][1] = conv4.bn
model.module_list[47][2] = conv4.act
spp = list(modelyolov5.model.children())[8]
model.module_list[48][0] = spp.cv1.conv
model.module_list[48][1] = spp.cv1.bn
model.module_list[48][2] = spp.cv1.act
model.module_list[49] = spp.m[0]
model.module_list[51] = spp.m[1]
model.module_list[53] = spp.m[2]
model.module_list[55][0] = spp.cv2.conv
model.module_list[55][1] = spp.cv2.bn
model.module_list[55][2] = spp.cv2.act
cspnet4 = list(modelyolov5.model.children())[9]
model.module_list[56][0] = cspnet4.cv2
model.module_list[58][0] = cspnet4.cv1.conv
model.module_list[58][1] = cspnet4.cv1.bn
model.module_list[58][2] = cspnet4.cv1.act
model.module_list[61][0] = cspnet4.cv3
model.module_list[63][0] = cspnet4.bn
model.module_list[63][1] = cspnet4.act
model.module_list[64][0] = cspnet4.cv4.conv
model.module_list[64][1] = cspnet4.cv4.bn
model.module_list[64][2] = cspnet4.cv4.act
model.module_list[59][0] = cspnet4.m[0].cv1.conv
model.module_list[59][1] = cspnet4.m[0].cv1.bn
model.module_list[59][2] = cspnet4.m[0].cv1.act
model.module_list[60][0] = cspnet4.m[0].cv2.conv
model.module_list[60][1] = cspnet4.m[0].cv2.bn
model.module_list[60][2] = cspnet4.m[0].cv2.act
conv5 = list(modelyolov5.model.children())[10]
model.module_list[65][0] = conv5.conv
model.module_list[65][1] = conv5.bn
model.module_list[65][2] = conv5.act
upsample1 = list(modelyolov5.model.children())[11]
model.module_list[66] = upsample1
cspnet5 = list(modelyolov5.model.children())[13]
model.module_list[68][0] = cspnet5.cv2
model.module_list[70][0] = cspnet5.cv1.conv
model.module_list[70][1] = cspnet5.cv1.bn
model.module_list[70][2] = cspnet5.cv1.act
model.module_list[73][0] = cspnet5.cv3
model.module_list[75][0] = cspnet5.bn
model.module_list[75][1] = cspnet5.act
model.module_list[76][0] = cspnet5.cv4.conv
model.module_list[76][1] = cspnet5.cv4.bn
model.module_list[76][2] = cspnet5.cv4.act
model.module_list[71][0] = cspnet5.m[0].cv1.conv
model.module_list[71][1] = cspnet5.m[0].cv1.bn
model.module_list[71][2] = cspnet5.m[0].cv1.act
model.module_list[72][0] = cspnet5.m[0].cv2.conv
model.module_list[72][1] = cspnet5.m[0].cv2.bn
model.module_list[72][2] = cspnet5.m[0].cv2.act
conv6 = list(modelyolov5.model.children())[14]
model.module_list[77][0] = conv6.conv
model.module_list[77][1] = conv6.bn
model.module_list[77][2] = conv6.act
upsample2 = list(modelyolov5.model.children())[15]
model.module_list[78] = upsample2
cspnet6 = list(modelyolov5.model.children())[17]
model.module_list[80][0] = cspnet6.cv2
model.module_list[82][0] = cspnet6.cv1.conv
model.module_list[82][1] = cspnet6.cv1.bn
model.module_list[82][2] = cspnet6.cv1.act
model.module_list[85][0] = cspnet6.cv3
model.module_list[87][0] = cspnet6.bn
model.module_list[87][1] = cspnet6.act
model.module_list[88][0] = cspnet6.cv4.conv
model.module_list[88][1] = cspnet6.cv4.bn
model.module_list[88][2] = cspnet6.cv4.act
model.module_list[83][0] = cspnet6.m[0].cv1.conv
model.module_list[83][1] = cspnet6.m[0].cv1.bn
model.module_list[83][2] = cspnet6.m[0].cv1.act
model.module_list[84][0] = cspnet6.m[0].cv2.conv
model.module_list[84][1] = cspnet6.m[0].cv2.bn
model.module_list[84][2] = cspnet6.m[0].cv2.act
conv7 = list(modelyolov5.model.children())[18]
model.module_list[92][0] = conv7.conv
model.module_list[92][1] = conv7.bn
model.module_list[92][2] = conv7.act
cspnet7 = list(modelyolov5.model.children())[20]
model.module_list[94][0] = cspnet7.cv2
model.module_list[96][0] = cspnet7.cv1.conv
model.module_list[96][1] = cspnet7.cv1.bn
model.module_list[96][2] = cspnet7.cv1.act
model.module_list[99][0] = cspnet7.cv3
model.module_list[101][0] = cspnet7.bn
model.module_list[101][1] = cspnet7.act
model.module_list[102][0] = cspnet7.cv4.conv
model.module_list[102][1] = cspnet7.cv4.bn
model.module_list[102][2] = cspnet7.cv4.act
model.module_list[97][0] = cspnet7.m[0].cv1.conv
model.module_list[97][1] = cspnet7.m[0].cv1.bn
model.module_list[97][2] = cspnet7.m[0].cv1.act
model.module_list[98][0] = cspnet7.m[0].cv2.conv
model.module_list[98][1] = cspnet7.m[0].cv2.bn
model.module_list[98][2] = cspnet7.m[0].cv2.act
conv8 = list(modelyolov5.model.children())[21]
model.module_list[106][0] = conv8.conv
model.module_list[106][1] = conv8.bn
model.module_list[106][2] = conv8.act
cspnet8 = list(modelyolov5.model.children())[23]
model.module_list[108][0] = cspnet8.cv2
model.module_list[110][0] = cspnet8.cv1.conv
model.module_list[110][1] = cspnet8.cv1.bn
model.module_list[110][2] = cspnet8.cv1.act
model.module_list[113][0] = cspnet8.cv3
model.module_list[115][0] = cspnet8.bn
model.module_list[115][1] = cspnet8.act
model.module_list[116][0] = cspnet8.cv4.conv
model.module_list[116][1] = cspnet8.cv4.bn
model.module_list[116][2] = cspnet8.cv4.act
model.module_list[111][0] = cspnet8.m[0].cv1.conv
model.module_list[111][1] = cspnet8.m[0].cv1.bn
model.module_list[111][2] = cspnet8.m[0].cv1.act
model.module_list[112][0] = cspnet8.m[0].cv2.conv
model.module_list[112][1] = cspnet8.m[0].cv2.bn
model.module_list[112][2] = cspnet8.m[0].cv2.act
detect = list(modelyolov5.model.children())[24]
model.module_list[89][0] = detect.m[0]
model.module_list[103][0] = detect.m[1]
model.module_list[117][0] = detect.m[2]
if __name__ == '__main__':
parser = argparse.ArgumentParser()
parser.add_argument('--cfg', type=str, default='cfg/yolov5s_v4.cfg', help='cfg file path')
parser.add_argument('--data', type=str, default='data/coco_128img.data', help='*.data file path')
parser.add_argument('--weights', type=str, default='weights/yolov5s_v4.pt', help='sparse model weights')
parser.add_argument('--percent', type=float, default=0.3, help='channel prune percent')
parser.add_argument('--img_size', type=int, default=416, help='inference size (pixels)')
opt = parser.parse_args()
print(opt)
img_size = opt.img_size
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
model = Darknet(opt.cfg, (img_size, img_size)).to(device)
modelyolov5 = torch.load(opt.weights, map_location=device)['model'].float() # load FP32 model
YOLOV5_V4 = True
if YOLOV5_V4:
# yolov5-v4
copy_weight_v4(modelyolov5, model)
else:
# yolov5-v3 yolov5-v2
copy_weight(modelyolov5, model)
eval_model = lambda model:test(model=model,cfg=opt.cfg, data=opt.data, batch_size=16, img_size=img_size)
obtain_num_parameters = lambda model:sum([param.nelement() for param in model.parameters()])
print("\nlet's test the original model first:")
with torch.no_grad():
origin_model_metric = eval_model(model)
origin_nparameters = obtain_num_parameters(model)
CBL_idx, Conv_idx, prune_idx,shortcut_idx,shortcut_all= parse_module_defs2(model.module_defs)
sort_prune_idx=[idx for idx in prune_idx if idx not in shortcut_idx]
#将所有要剪枝的BN层的α参数，拷贝到bn_weights列表
bn_weights = gather_bn_weights(model.module_list, sort_prune_idx)
#torch.sort返回二维列表，第一维是排序后的值列表，第二维是排序后的值列表对应的索引
sorted_bn = torch.sort(bn_weights)[0]
#避免剪掉所有channel的最高阈值(每个BN层的gamma的最大值的最小值即为阈值上限)
highest_thre = []
for idx in sort_prune_idx:
#.item()可以得到张量里的元素值
# highest_thre.append(model.module_list[idx][1].weight.data.abs().max().item())
highest_thre.append(model.module_list[idx][1].weight.data.abs().max().item() if type(
model.module_list[idx][1]).__name__ is 'BatchNorm2d' else model.module_list[idx][
0].weight.data.abs().max().item())
highest_thre = min(highest_thre)
# 找到highest_thre对应的下标对应的百分比
percent_limit = (sorted_bn==highest_thre).nonzero().item()/len(bn_weights)
print(f'Suggested Threshold should be less than {highest_thre:.4f}.')
print(f'The corresponding prune ratio is {percent_limit:.3f},but you can set higher.')
def prune_and_eval(model, sorted_bn, percent=.0):
model_copy = deepcopy(model)
thre_index = int(len(sorted_bn) * percent)
#获得α参数的阈值，小于该值的α参数对应的通道，全部裁剪掉
thre1 = sorted_bn[thre_index]
print(f'Channels with Gamma value less than {thre1:.6f} are pruned!')
remain_num = 0
idx_new=dict()
for idx in prune_idx:
if idx not in shortcut_idx:
# bn_module = model_copy.module_list[idx][1]
bn_module = model_copy.module_list[idx][1] if type(
model_copy.module_list[idx][1]).__name__ is 'BatchNorm2d' else model_copy.module_list[idx][0]
mask = obtain_bn_mask(bn_module, thre1)
#记录剪枝后，每一层卷积层对应的mask
# idx_new[idx]=mask.cpu().numpy()
idx_new[idx]=mask
remain_num += int(mask.sum())
bn_module.weight.data.mul_(mask)
#bn_module.bias.data.mul_(mask*0.0001)
else:
bn_module = model_copy.module_list[idx][1]
mask=idx_new[shortcut_idx[idx]]
idx_new[idx]=mask
remain_num += int(mask.sum())
bn_module.weight.data.mul_(mask)
#print(int(mask.sum()))
with torch.no_grad():
mAP = eval_model(model_copy)[0][2]
print(f'Number of channels has been reduced from {len(sorted_bn)} to {remain_num}')
print(f'Prune ratio: {1-remain_num/len(sorted_bn):.3f}')
print(f'mAP of the pruned model is {mAP:.4f}')
return thre1
percent = opt.percent
threshold = prune_and_eval(model, sorted_bn, percent)
#****************************************************************
#虽然上面已经能看到剪枝后的效果，但是没有生成剪枝后的模型结构，因此下面的代码是为了生成新的模型结构并拷贝旧模型参数到新模型
#%%
def obtain_filters_mask(model, thre, CBL_idx, prune_idx):
pruned = 0
total = 0
num_filters = []
filters_mask = []
idx_new=dict()
#CBL_idx存储的是所有带BN的卷积层（YOLO层的前一层卷积层是不带BN的）
for idx in CBL_idx:
# bn_module = model.module_list[idx][1]
bn_module = model.module_list[idx][1] if type(model.module_list[idx][1]).__name__ is 'BatchNorm2d' else \
model.module_list[idx][0]
if idx in prune_idx:
if idx not in shortcut_idx:
mask = obtain_bn_mask(bn_module, thre).cpu().numpy()
if type(model.module_list[idx][0]).__name__ is 'BatchNorm2d':
half_num = int(len(mask) / 2)
mask1 = mask[:half_num]
mask2 = mask[half_num:]
remain1 = int(mask1.sum())
remain2 = int(mask2.sum())
if remain1 == 0 or remain2 == 0:
print("Channels would be all pruned!")
raise Exception
idx_new[idx]=mask
remain = int(mask.sum())
pruned = pruned + mask.shape[0] - remain
# if remain == 0:
# print("Channels would be all pruned!")
# raise Exception
# print(f'layer index: {idx:>3d} \t total channel: {mask.shape[0]:>4d} \t '
# f'remaining channel: {remain:>4d}')
else:
mask=idx_new[shortcut_idx[idx]]
idx_new[idx]=mask
remain= int(mask.sum())
pruned = pruned + mask.shape[0] - remain
if remain == 0:
# print("Channels would be all pruned!")
# raise Exception
max_value = bn_module.weight.data.abs().max()
mask = obtain_bn_mask(bn_module, max_value).cpu().numpy()
remain = int(mask.sum())
pruned = pruned + mask.shape[0] - remain
print(f'layer index: {idx:>3d} \t total channel: {mask.shape[0]:>4d} \t '
f'remaining channel: {remain:>4d}')
else:
mask = np.ones(bn_module.weight.data.shape)
remain = mask.shape[0]
total += mask.shape[0]
num_filters.append(remain)
filters_mask.append(mask.copy())
#因此，这里求出的prune_ratio,需要裁剪的α参数/cbl_idx中所有的α参数
prune_ratio = pruned / total
print(f'Prune channels: {pruned}\tPrune ratio: {prune_ratio:.3f}')
return num_filters, filters_mask
num_filters, filters_mask = obtain_filters_mask(model, threshold, CBL_idx, prune_idx)
#CBLidx2mask存储CBL_idx中，每一层BN层对应的mask
CBLidx2mask = {idx: mask for idx, mask in zip(CBL_idx, filters_mask)}
pruned_model = prune_model_keep_size2(model, prune_idx, CBL_idx, CBLidx2mask)
print("\nnow prune the model but keep size,(actually add offset of BN beta to next layer), let's see how the mAP goes")
with torch.no_grad():
eval_model(pruned_model)
#获得原始模型的module_defs，并修改该defs中的卷积核数量
compact_module_defs = deepcopy(model.module_defs)
for idx, num in zip(CBL_idx, num_filters):
assert compact_module_defs[idx]['type'] == 'convolutional' or compact_module_defs[idx]['type'] == 'convolutional_noconv'
compact_module_defs[idx]['filters'] = str(num)
if compact_module_defs[idx]['type'] == 'convolutional_noconv':
model_def=compact_module_defs[idx-1] #route
assert compact_module_defs[idx-1]['type'] == 'route'
from_layers = [int(s) for s in model_def['layers'].split(',')]
assert compact_module_defs[idx - 1 + from_layers[0]]['type'] == 'convolutional_nobias'
assert compact_module_defs[idx-1 + from_layers[1] if from_layers[1] < 0 else from_layers[1]]['type'] == 'convolutional_nobias'
half_num = int(len(CBLidx2mask[idx]) / 2)
mask1=CBLidx2mask[idx][:half_num]
mask2 = CBLidx2mask[idx][half_num:]
remain1 = int(mask1.sum())
remain2 = int(mask2.sum())
compact_module_defs[idx - 1 + from_layers[0]]['filters']=remain1
compact_module_defs[idx-1 + from_layers[1] if from_layers[1] < 0 else from_layers[1]]['filters'] =remain2
compact_model = Darknet([model.hyperparams.copy()] + compact_module_defs, (img_size, img_size)).to(device)
compact_nparameters = obtain_num_parameters(compact_model)
init_weights_from_loose_model(compact_model, pruned_model, CBL_idx, Conv_idx, CBLidx2mask)
random_input = torch.rand((1, 3, img_size, img_size)).to(device)
def obtain_avg_forward_time(input, model, repeat=200):
model.eval()
start = time.time()
with torch.no_grad():
for i in range(repeat):
output = model(input)[0]
avg_infer_time = (time.time() - start) / repeat
return avg_infer_time, output
print('testing Inference time...')
pruned_forward_time, pruned_output = obtain_avg_forward_time(random_input, pruned_model)
compact_forward_time, compact_output = obtain_avg_forward_time(random_input, compact_model)
diff = (pruned_output - compact_output).abs().gt(0.001).sum().item()
if diff > 0:
print('Something wrong with the pruned model!')
# 在测试集上测试剪枝后的模型, 并统计模型的参数数量
print('testing final model')
with torch.no_grad():
compact_model_metric = eval_model(compact_model)
# 比较剪枝前后参数数量的变化、指标性能的变化
metric_table = [
["Metric", "Before", "After"],
["mAP", f'{origin_model_metric[0][2]:.6f}', f'{compact_model_metric[0][2]:.6f}'],
["Parameters", f"{origin_nparameters}", f"{compact_nparameters}"],
["Inference", f'{pruned_forward_time:.4f}', f'{compact_forward_time:.4f}']
]
print(AsciiTable(metric_table).table)
# 生成剪枝后的cfg文件并保存模型
pruned_cfg_name = opt.cfg.replace('/', f'/prune_{percent}_')
pruned_cfg_file = write_cfg(pruned_cfg_name, [model.hyperparams.copy()] + compact_module_defs)
print(f'Config file has been saved: {pruned_cfg_file}')
compact_model_name = opt.weights.replace('/', f'/prune_{percent}_')
if compact_model_name.endswith('.pt'):
chkpt = {'epoch': -1,
'best_fitness': None,
'training_results': None,
'model': compact_model.state_dict(),
'optimizer': None}
torch.save(chkpt, compact_model_name)
compact_model_name = compact_model_name.replace('.pt', '.weights')
# save_weights(compact_model, path=compact_model_name)
print(f'Compact model has been saved: {compact_model_name}')