-
Notifications
You must be signed in to change notification settings - Fork 0
/
test_net_wordvec_split.py
366 lines (311 loc) · 22.8 KB
/
test_net_wordvec_split.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
# --------------------------------------------------------
# Tensorflow Faster R-CNN
# Licensed under The MIT License [see LICENSE for details]
# Written by Jiasen Lu, Jianwei Yang, based on code from Ross Girshick
# --------------------------------------------------------
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
import _init_paths
import os
import sys
import numpy as np
import argparse
import pprint
import pdb
import time
import cv2
import torch
from torch.autograd import Variable
import torch.nn as nn
import torch.optim as optim
import pickle
from roi_data_layer.roidb import combined_roidb
from roi_data_layer.roibatchLoader import roibatchLoader
from model.utils.config import cfg, cfg_from_file, cfg_from_list, get_output_dir
from model.rpn.bbox_transform import clip_boxes
from model.nms.nms_wrapper import nms
from model.rpn.bbox_transform import bbox_transform_inv
from model.utils.net_utils import save_net, load_net, vis_detections
from model.faster_rcnn.vgg16 import vgg16
from model.faster_rcnn.resnet_w2v import resnet
import pdb
try:
xrange # Python 2
except NameError:
xrange = range # Python 3
def cosine_similarity(vec1, vec2):
return np.dot(vec1, vec2)/(np.linalg.norm(vec1)* np.linalg.norm(vec2) + 1e-8)
name_vectors = {0:[float(x) for x in "-0.6742 0.85137 -0.44563 -0.13542 0.66321 0.40091 -0.17317 -0.31843 -0.50706 -0.16611 -0.0037329 -0.15644 0.49711 0.42111 -0.15725 -0.29234 -0.49116 -0.41583 0.40411 -0.98499 0.038447 0.58983 -0.097437 0.13938 0.21717 -0.79006 -1.1404 0.063862 0.091416 0.13035 2.6491 -0.39185 0.10433 -1.2945 0.25436 0.68762 -0.24406 -0.55442 -0.26576 -0.19875 0.4121 0.48016 0.10084 -0.0076807 0.14454 -0.2255 0.78955 0.49908 0.1241 0.06661".split()],
1:[float(x) for x in "0.27203 -0.83471 -0.77436 -0.52342 -0.7846 0.43548 -0.55723 -0.12461 0.40705 -0.26957 0.9665 0.65333 -0.16101 -0.12138 -1.2511 -0.3532 -0.99238 1.3659 -1.4554 -0.61021 0.2708 0.23935 -0.68575 -0.38805 0.31965 -0.43329 -0.56755 0.32242 0.7346 0.21644 0.022579 -0.34303 -0.19963 1.0519 0.80017 -0.27862 0.70645 0.18759 -0.46561 0.033631 0.9176 -1.2649 0.23333 0.1328 0.32079 0.038267 0.2414 0.0043794 0.046835 0.20238".split()],
2:[float(x) for x in "-0.41716 0.16271 0.18208 0.040874 -0.76743 0.47832 -0.95683 -0.69713 1.3446 -0.59146 0.49005 -0.59504 -0.96514 0.41526 -0.053253 -0.65851 -0.33963 1.4676 -0.51452 -1.5919 0.44235 -0.10577 -1.4277 1.1811 0.2842 -0.92328 -0.2138 0.10129 0.61985 -0.21404 1.3418 0.53977 -0.65476 0.55278 0.38827 0.34769 0.37724 0.45527 -0.30526 0.97048 -0.016631 -0.36535 0.060563 0.33608 0.098575 -0.96056 0.31872 -1.1316 0.91703 -0.65762".split()],
3:[float(x) for x in "0.78675 0.079368 -0.76597 0.1931 0.55014 0.26493 -0.75841 -0.8818 1.6468 -0.54381 0.33519 0.44399 1.089 0.27044 0.74471 0.2487 0.2491 -0.28966 -1.4556 0.35605 -1.1725 -0.49858 0.35345 -0.1418 0.71734 -1.1416 -0.038701 0.27515 -0.017704 -0.44013 1.9597 -0.064666 0.47177 -0.03 -0.31617 0.26984 0.56195 -0.27882 -0.36358 -0.21923 -0.75046 0.31817 0.29354 0.25109 1.6111 0.7134 -0.15243 -0.25362 0.26419 0.15875".split()],
4:[float(x) for x in "0.96989 0.075746 0.157 -0.10599 -0.6147 0.085639 -1.3175 0.63774 0.84952 -1.0661 0.023076 0.69078 0.49954 0.89659 -0.11993 -0.68484 -0.055934 0.89771 -1.6127 -0.93314 0.34222 0.56598 -0.49005 0.064676 1.27 -1.5572 0.47292 0.36106 0.58891 -0.25283 2.1768 0.029947 -0.45634 1.1527 0.38669 0.26368 0.32151 -0.50073 -0.34753 0.094688 -0.10259 -0.34041 1.2957 0.019886 0.056735 -0.68682 -0.24096 -0.91495 0.73288 -0.95561".split()],
5:[float(x) for x in "0.20635 0.2606 -0.094723 -0.73396 0.72598 0.5099 -0.39352 -0.45703 0.49335 1.3791 0.10285 0.14997 0.41506 -0.19039 1.0527 0.16514 -0.16717 0.8092 -0.97394 -1.753 0.34632 -0.053064 0.33046 -0.021036 -0.78655 -1.0088 -0.30341 1.6766 0.90808 -0.39309 1.2131 0.21588 -0.87778 1.3756 0.57432 0.35111 0.39926 0.33184 1.2035 -0.21218 1.2316 0.58557 -0.40531 0.37376 0.16584 0.56948 -0.13898 -0.29062 0.56082 -0.94112".split()],
6:[float(x) for x in "0.84772 0.070253 0.96791 -0.27164 -0.37617 0.31978 -1.3108 0.091093 0.59919 -0.90217 -0.050876 -0.83886 -0.61596 0.29642 -0.42189 -0.21969 -0.94006 1.2221 -0.66526 -0.57745 0.76126 0.51459 -0.88565 1.5135 0.42326 -1.2947 0.45522 0.67073 0.80188 -0.65449 2.4117 0.62445 -0.046631 0.37524 1.0103 0.25259 1.0913 -0.79427 -0.17027 1.4866 -0.24077 0.021904 -0.01603 -0.44319 -0.13914 0.013311 -0.49432 -0.57696 1.1997 -0.39707".split()],
7:[float(x) for x in "0.47685 -0.084552 1.4641 0.047017 0.14686 0.5082 -1.2228 -0.22607 0.19306 -0.29756 0.20599 -0.71284 -1.6288 0.17096 0.74797 -0.061943 -0.65766 1.3786 -0.68043 -1.7551 0.58319 0.25157 -1.2114 0.81343 0.094825 -1.6819 -0.64498 0.6322 1.1211 0.16112 2.5379 0.24852 -0.26816 0.32818 1.2916 0.23548 0.61465 -0.1344 -0.13237 0.27398 -0.11821 0.1354 0.074306 -0.61951 0.45472 -0.30318 -0.21883 -0.56054 1.1177 -0.36595".split()],
8:[float(x) for x in "0.45281 -0.50108 -0.53714 -0.015697 0.22191 0.54602 -0.67301 -0.6891 0.63493 -0.19726 0.33685 0.7735 0.90094 0.38488 0.38367 0.2657 -0.08057 0.61089 -1.2894 -0.22313 -0.61578 0.21697 0.35614 0.44499 0.60885 -1.1633 -1.1579 0.36118 0.10466 -0.78325 1.4352 0.18629 -0.26112 0.83275 -0.23123 0.32481 0.14485 -0.44552 0.33497 -0.95946 -0.097479 0.48138 -0.43352 0.69455 0.91043 -0.28173 0.41637 -1.2609 0.71278 0.23782".split()],
9:[float(x) for x in "-1.0443 0.49202 -0.75978 -0.39224 0.81217 -0.039287 0.016706 -0.68629 -0.078359 -1.3214 -0.15354 0.20438 -0.46503 1.2145 -0.18217 0.27451 -0.24086 0.71145 0.3247 -0.7132 0.66721 0.71307 -0.10394 -0.38439 -0.2026 -1.4419 0.42644 0.59436 -1.3615 0.0013784 1.8734 -0.11334 -0.88115 -0.21715 -0.56606 0.14152 0.27673 0.99962 1.0567 -0.29428 -0.3139 0.12729 -0.54363 0.39652 -0.32527 0.30536 0.15128 -1.0889 -0.20867 -0.052605".split()],
10:[float(x) for x in "0.61253 -0.48167 -0.74199 -0.55203 -0.007596 1.6101 -0.88565 -0.81981 1.5144 -0.22804 0.55367 -0.18392 0.7049 -0.36931 1.0668 1.1077 0.19709 0.24731 -0.68395 0.5475 -0.038255 -0.78989 0.61131 0.31473 0.50215 -1.6535 -0.42782 1.0404 0.29429 -0.36889 1.3148 -0.18443 0.092753 0.77572 -0.54845 -0.14645 0.51128 0.047248 0.41781 -0.18324 -0.44197 -0.25237 -0.3359 0.3096 1.9192 0.3396 -0.27341 -0.01316 0.64974 -0.85857".split()],
11:[float(x) for x in "0.08052 2.4489 -1.82518 -0.0477 2.00389 -1.69875 -0.751511 -0.50848 -0.428588 -1.31098 -1.68757 -0.32697 0.07998 1.65463 0.90282 0.26249 0.13324 0.62698 0.371472 -2.6305 1.60213 0.56334 -0.38732 1.69456 -0.58381 -0.632158 -0.41 1.51932 -0.09594 0.21045 4.788 0.99336 -0.44077 0.63905 0.6379 1.42181 0.47585 1.95477 0.327167 -0.12653 0.77951 -0.10258 -0.179286 2.03756 0.1777877 0.9459 0.39777 -0.778347 0.99945 -0.651".split()],
12:[float(x) for x in "0.11008 -0.38781 -0.57615 -0.27714 0.70521 0.53994 -1.0786 -0.40146 1.1504 -0.5678 0.0038977 0.52878 0.64561 0.47262 0.48549 -0.18407 0.1801 0.91397 -1.1979 -0.5778 -0.37985 0.33606 0.772 0.75555 0.45506 -1.7671 -1.0503 0.42566 0.41893 -0.68327 1.5673 0.27685 -0.61708 0.64638 -0.076996 0.37118 0.1308 -0.45137 0.25398 -0.74392 -0.086199 0.24068 -0.64819 0.83549 1.2502 -0.51379 0.04224 -0.88118 0.7158 0.38519".split()],
13:[float(x) for x in "-0.20454 0.23321 -0.59158 -0.29205 0.29391 0.31169 -0.94937 0.055974 1.0031 -1.0761 -0.0094648 0.18381 -0.048405 -0.35717 0.26004 -0.41028 0.51489 1.2009 -1.6136 -1.1003 -0.23455 -0.81654 -0.15103 0.37068 0.477 -1.7027 -1.2183 0.038898 0.23327 0.028245 1.6588 0.26703 -0.29938 0.99149 0.34263 0.15477 0.028372 0.56276 -0.62823 -0.67923 -0.163 -0.49922 -0.8599 0.85469 0.75059 -1.0399 -0.11033 -1.4237 0.65984 -0.3198".split()],
14:[float(x) for x in "0.14362 -1.1402 0.39368 0.18135 -0.094088 0.67473 -0.52618 0.21466 0.62416 -0.17217 0.67109 -1.1389 -0.84819 0.085305 0.20975 -0.59836 -0.78554 1.21 -0.90412 -1.009 0.42731 0.39614 -1.0663 0.66758 0.54771 -0.93963 -0.31805 0.14893 0.4489 -0.1986 0.20147 0.47226 -0.31627 0.83248 0.84036 0.40339 0.24902 -0.034884 -0.11794 0.89527 -0.33927 0.13761 -0.037933 -0.26963 0.85965 -1.174 0.31216 -0.62433 1.4447 -1.0968".split()],
15:[float(x) for x in "0.61734 0.40035 0.067786 -0.34263 2.0647 0.60844 0.32558 0.3869 0.36906 0.16553 0.0065053 -0.075674 0.57099 0.17314 1.0142 -0.49581 -0.38152 0.49255 -0.16737 -0.33948 -0.44405 0.77543 0.20935 0.6007 0.86649 -1.8923 -0.37901 -0.28044 0.64214 -0.23549 2.9358 -0.086004 -0.14327 -0.50161 0.25291 -0.065446 0.60768 0.13984 0.018135 -0.34877 0.039985 0.07943 0.39318 1.0562 -0.23624 -0.4194 -0.35332 -0.15234 0.62158 0.79257".split()],
16:[float(x) for x in "1.87558 0.70306 -0.13368 1.0749 0.344 0.359 -1.29974 -0.81254 0.38347 0.20391 0.19335 -0.2915 1.75308 -1.23573 1.04908 0.20029 0.01158 2.52851 -0.17232 -1.70577 -0.50728 -2.48107 -0.60158 -1.52139 0.59061 -0.46542 1.71796 1.25714 1.44733 -0.97309 2.35745 -1.07286 -0.67348 -0.21713 0.02439 0.835637 -0.481641 2.17821 1.30031 0.16752 -0.37618 -1.94368 0.213935 -0.00917 2.51475 1.63038 -0.311639 -1.106158 -0.94201 -0.58308".split()],
17:[float(x) for x in "0.39026 -1.1357 -0.47646 -1.0185 0.44893 1.4499 -1.5251 -0.97275 1.5137 -0.58559 0.68279 -0.63901 1.6961 -1.233 0.50333 0.6791 1.0007 0.53626 -1.1749 -0.054361 -0.42404 -0.49371 0.43183 0.40727 0.74916 -0.53304 -0.33745 0.11388 0.47025 -0.47616 1.3084 0.1534 0.09754 1.0811 -0.20525 0.14341 -0.080796 0.17984 0.12433 0.12709 -0.31428 0.035969 -0.4609 0.53133 1.4986 -0.22609 -0.46637 -0.54343 -0.27696 -0.75601".split()],
18:[float(x) for x in "-0.12335 0.23509 -0.7438 -0.50219 0.97023 -0.12771 0.039666 -0.67514 -0.2298 -1.2603 -0.18912 0.019617 -0.096285 1.3191 0.31878 1.3884 -0.41529 0.33527 0.94093 -0.50369 0.37852 1.104 -0.41287 -0.10135 -0.59901 0.060218 -0.21657 1.5896 0.57137 0.061694 0.66554 0.39239 -0.73251 1.401 -0.095607 0.66528 0.36132 -0.014117 0.068288 -0.62261 -0.023754 -0.087556 0.1543 0.5651 -1.0085 0.065175 0.2977 -1.0227 -0.21718 -0.49374".split()],
19:[float(x) for x in "0.94971 0.34328 0.84504 -0.88519 -0.72078 -0.29309 -0.74678 0.65122 0.47295 -0.74011 0.1877 -0.38279 -0.55899 0.42952 -0.26984 -0.42383 -0.31236 1.3423 -0.78567 -0.6302 0.91819 0.21126 -0.57442 1.4549 0.75456 -1.6165 -0.0085015 0.0029134 0.51304 -0.47447 2.5306 0.85944 -0.30667 0.057765 0.66231 0.20804 0.64237 -0.5246 -0.053416 1.1404 -0.13703 -0.18361 0.45459 -0.50963 -0.025539 -0.02861 0.18048 -0.4483 0.40525 -0.36821".split()],
20:[float(x) for x in "0.25626 -0.17277 1.52866 1.06081 -0.56051 -1.009539 -0.845252 -1.18025 1.38689 -0.24081 0.622749 0.121095 0.02118 2.1764 0.712973 1.2033 -0.87334 0.5934 -0.11 -0.34415 1.34936 1.39815 1.1527 1.63682 0.0212764 -2.23525 -0.977023 0.09278 -0.01013 -0.06802 5.4253 0.57827 -0.6189 -1.3413 -1.44589 0.63279 0.43006 -1.66446 -0.96338 -0.87314 0.99411 1.098152 -0.24369 -0.4259232 -0.697979 -0.101647 -0.04148 0.58366 -0.26210906 0.947361".split()]}
def parse_args():
"""
Parse input arguments
"""
parser = argparse.ArgumentParser(description='Train a Fast R-CNN network')
parser.add_argument('--dataset', dest='dataset',
help='training dataset',
default='pascal_voc', type=str)
parser.add_argument('--cfg', dest='cfg_file',
help='optional config file',
default='cfgs/vgg16.yml', type=str)
parser.add_argument('--net', dest='net',
help='vgg16, res50, res101, res152',
default='res101', type=str)
parser.add_argument('--set', dest='set_cfgs',
help='set config keys', default=None,
nargs=argparse.REMAINDER)
parser.add_argument('--load_dir', dest='load_dir',
help='directory to load models', default="models",
type=str)
parser.add_argument('--cuda', dest='cuda',
help='whether use CUDA',
action='store_true')
parser.add_argument('--ls', dest='large_scale',
help='whether use large imag scale',
action='store_true')
parser.add_argument('--mGPUs', dest='mGPUs',
help='whether use multiple GPUs',
action='store_true')
parser.add_argument('--cag', dest='class_agnostic',
help='whether perform class_agnostic bbox regression',
action='store_true')
parser.add_argument('--parallel_type', dest='parallel_type',
help='which part of model to parallel, 0: all, 1: model before roi pooling',
default=0, type=int)
parser.add_argument('--checksession', dest='checksession',
help='checksession to load model',
default=1, type=int)
parser.add_argument('--checkepoch', dest='checkepoch',
help='checkepoch to load network',
default=1, type=int)
parser.add_argument('--checkpoint', dest='checkpoint',
help='checkpoint to load network',
default=10021, type=int)
parser.add_argument('--vis', dest='vis',
help='visualization mode',
action='store_true')
parser.add_argument('--gpus', nargs='+', type=int, default=None)
args = parser.parse_args()
return args
lr = cfg.TRAIN.LEARNING_RATE
momentum = cfg.TRAIN.MOMENTUM
weight_decay = cfg.TRAIN.WEIGHT_DECAY
if __name__ == '__main__':
args = parse_args()
print('Called with args:')
print(args)
if torch.cuda.is_available() and not args.cuda:
print("WARNING: You have a CUDA device, so you should probably run with --cuda")
np.random.seed(cfg.RNG_SEED)
if args.dataset == "pascal_voc":
args.imdb_name = "voc_2007_trainval"
args.imdbval_name = "voc_2007_test"
args.set_cfgs = ['ANCHOR_SCALES', '[8, 16, 32]', 'ANCHOR_RATIOS', '[0.5,1,2]']
elif args.dataset == "pascal_voc_0712":
args.imdb_name = "voc_2007_trainval+voc_2012_trainval"
args.imdbval_name = "voc_2007_test"
args.set_cfgs = ['ANCHOR_SCALES', '[8, 16, 32]', 'ANCHOR_RATIOS', '[0.5,1,2]']
elif args.dataset == "coco":
args.imdb_name = "coco_2014_train+coco_2014_valminusminival"
args.imdbval_name = "coco_2014_minival"
args.set_cfgs = ['ANCHOR_SCALES', '[4, 8, 16, 32]', 'ANCHOR_RATIOS', '[0.5,1,2]']
elif args.dataset == "imagenet":
args.imdb_name = "imagenet_train"
args.imdbval_name = "imagenet_val"
args.set_cfgs = ['ANCHOR_SCALES', '[8, 16, 32]', 'ANCHOR_RATIOS', '[0.5,1,2]']
elif args.dataset == "vg":
args.imdb_name = "vg_150-50-50_minitrain"
args.imdbval_name = "vg_150-50-50_minival"
args.set_cfgs = ['ANCHOR_SCALES', '[4, 8, 16, 32]', 'ANCHOR_RATIOS', '[0.5,1,2]']
args.cfg_file = "cfgs/{}_ls.yml".format(args.net) if args.large_scale else "cfgs/{}.yml".format(args.net)
if args.cfg_file is not None:
cfg_from_file(args.cfg_file)
if args.set_cfgs is not None:
cfg_from_list(args.set_cfgs)
print('Using config:')
pprint.pprint(cfg)
cfg.TRAIN.USE_FLIPPED = False
imdb, roidb, ratio_list, ratio_index = combined_roidb(args.imdbval_name, False)
imdb.competition_mode(on=True)
print('{:d} roidb entries'.format(len(roidb)))
input_dir = args.load_dir + "/" + args.net + "/" + args.dataset
if not os.path.exists(input_dir):
raise Exception('There is no input directory for loading network from ' + input_dir)
load_name = os.path.join(input_dir,
'faster_rcnn_{}_{}_{}.pth'.format(args.checksession, args.checkepoch, args.checkpoint))
# initilize the network here.
if args.net == 'vgg16':
fasterRCNN = vgg16(imdb.classes, pretrained=False, class_agnostic=args.class_agnostic)
elif args.net == 'res101':
fasterRCNN = resnet(imdb.classes, 101, pretrained=False, class_agnostic=args.class_agnostic)
elif args.net == 'res50':
fasterRCNN = resnet(imdb.classes, 50, pretrained=False, class_agnostic=args.class_agnostic)
elif args.net == 'res152':
fasterRCNN = resnet(imdb.classes, 152, pretrained=False, class_agnostic=args.class_agnostic)
else:
print("network is not defined")
pdb.set_trace()
fasterRCNN.create_architecture()
print("load checkpoint %s" % (load_name))
#map_location={'cuda:0':'cuda:{}'.format(int(args.gpus[0]))}
checkpoint = torch.load(load_name)
fasterRCNN.load_state_dict(checkpoint['model'])
if 'pooling_mode' in checkpoint.keys():
cfg.POOLING_MODE = checkpoint['pooling_mode']
print('load model successfully!')
# initilize the tensor holder here.
im_data = torch.FloatTensor(1)
im_info = torch.FloatTensor(1)
num_boxes = torch.LongTensor(1)
gt_boxes = torch.FloatTensor(1)
# ship to cuda
if args.cuda:
im_data = im_data.cuda(device=args.gpus[0])
im_info = im_info.cuda(device=args.gpus[0])
num_boxes = num_boxes.cuda(device=args.gpus[0])
gt_boxes = gt_boxes.cuda(device=args.gpus[0])
# make variable
im_data = Variable(im_data)
im_info = Variable(im_info)
num_boxes = Variable(num_boxes)
gt_boxes = Variable(gt_boxes)
if args.cuda:
cfg.CUDA = True
if args.cuda:
fasterRCNN.cuda(device=args.gpus[0])
#fasterRCNN.to(torch.device("cuda:{}".format(args.gpus[0])))
start = time.time()
max_per_image = 100
vis = args.vis
if vis:
thresh = 0.05
else:
thresh = 0.0
save_name = 'faster_rcnn_{}_{}'.format(args.checksession, args.checkepoch)
num_images = len(imdb.image_index)
all_boxes = [[[] for _ in xrange(num_images)]
for _ in xrange(imdb.num_classes)]
output_dir = get_output_dir(imdb, save_name)
dataset = roibatchLoader(roidb, ratio_list, ratio_index, 1, \
imdb.num_classes, training=False, normalize = False)
dataloader = torch.utils.data.DataLoader(dataset, batch_size=1,
shuffle=False, num_workers=0,
pin_memory=True)
data_iter = iter(dataloader)
_t = {'im_detect': time.time(), 'misc': time.time()}
det_file = os.path.join(output_dir, 'detections.pkl')
fasterRCNN.eval()
empty_array = np.transpose(np.array([[],[],[],[],[]]), (1,0))
for i in range(num_images):
data = next(data_iter)
im_data.data.resize_(data[0].size()).copy_(data[0])
im_info.data.resize_(data[1].size()).copy_(data[1])
gt_boxes.data.resize_(data[2].size()).copy_(data[2])
num_boxes.data.resize_(data[3].size()).copy_(data[3])
det_tic = time.time()
rois, cls_prob, bbox_pred, \
rpn_loss_cls, rpn_loss_box, \
RCNN_loss_cls, RCNN_loss_bbox, \
rois_label = fasterRCNN(im_data, im_info, gt_boxes, num_boxes)
scores = cls_prob.data
boxes = rois.data[:, :, 1:5]
if cfg.TEST.BBOX_REG:
# Apply bounding-box regression deltas
box_deltas = bbox_pred.data
if cfg.TRAIN.BBOX_NORMALIZE_TARGETS_PRECOMPUTED:
# Optionally normalize targets by a precomputed mean and stdev
if args.class_agnostic:
box_deltas = box_deltas.view(-1, 4) * torch.FloatTensor(cfg.TRAIN.BBOX_NORMALIZE_STDS).cuda(device=args.gpus[0]) \
+ torch.FloatTensor(cfg.TRAIN.BBOX_NORMALIZE_MEANS).cuda(device=args.gpus[0])
box_deltas = box_deltas.view(1, -1, 4)
else:
box_deltas = box_deltas.view(-1, 4) * torch.FloatTensor(cfg.TRAIN.BBOX_NORMALIZE_STDS).cuda(device=args.gpus[0]) \
+ torch.FloatTensor(cfg.TRAIN.BBOX_NORMALIZE_MEANS).cuda(device=args.gpus[0])
box_deltas = box_deltas.view(1, -1, 4 * len(imdb.classes))
pred_boxes = bbox_transform_inv(boxes, box_deltas, 1)
pred_boxes = clip_boxes(pred_boxes, im_info.data, 1)
else:
# Simply repeat the boxes, once for each class
pred_boxes = np.tile(boxes, (1, scores.shape[1]))
pred_boxes /= data[1][0][2].item()
scores = scores.squeeze()
softmax = nn.Softmax(dim=1)
res_vector = []
for ii in range(scores.shape[0]): # for all 300 the bounding boxes of an image
res_vector.append([])
res_vector[ii].append(-1)
for k in range(1, 21): # we calculate the dot product with each possible class
res_vector[ii].append(cosine_similarity(scores.cpu().numpy()[ii], np.array(name_vectors[k])))
scores = softmax(torch.tensor(res_vector).cuda(device=args.gpus[0]))
pred_boxes = pred_boxes.squeeze()
det_toc = time.time()
detect_time = det_toc - det_tic
misc_tic = time.time()
if vis:
im = cv2.imread(imdb.image_path_at(i))
im2show = np.copy(im)
for j in xrange(1, imdb.num_classes):
inds = torch.nonzero(scores[:,j]>thresh).view(-1)
# if there is det
if inds.numel() > 0:
cls_scores = scores[:,j][inds]
_, order = torch.sort(cls_scores, 0, True)
if args.class_agnostic:
cls_boxes = pred_boxes[inds, :]
else:
cls_boxes = pred_boxes[inds][:, j * 4:(j + 1) * 4]
cls_dets = torch.cat((cls_boxes, cls_scores.unsqueeze(1)), 1)
# cls_dets = torch.cat((cls_boxes, cls_scores), 1)
cls_dets = cls_dets[order]
keep = nms(cls_dets, cfg.TEST.NMS)
cls_dets = cls_dets[keep.view(-1).long()]
if vis:
im2show = vis_detections(im2show, imdb.classes[j], cls_dets.cpu().numpy(), 0.3)
all_boxes[j][i] = cls_dets.cpu().numpy()
else:
all_boxes[j][i] = empty_array
# Limit to max_per_image detections *over all classes*
if max_per_image > 0:
image_scores = np.hstack([all_boxes[j][i][:, -1]
for j in xrange(1, imdb.num_classes)])
if len(image_scores) > max_per_image:
image_thresh = np.sort(image_scores)[-max_per_image]
for j in xrange(1, imdb.num_classes):
keep = np.where(all_boxes[j][i][:, -1] >= image_thresh)[0]
all_boxes[j][i] = all_boxes[j][i][keep, :]
misc_toc = time.time()
nms_time = misc_toc - misc_tic
sys.stdout.write('im_detect: {:d}/{:d} {:.3f}s {:.3f}s \r' \
.format(i + 1, num_images, detect_time, nms_time))
sys.stdout.flush()
if vis:
cv2.imwrite('result.png', im2show)
pdb.set_trace()
#cv2.imshow('test', im2show)
#cv2.waitKey(0)
with open(det_file, 'wb') as f:
pickle.dump(all_boxes, f, pickle.HIGHEST_PROTOCOL)
print('Evaluating detections')
imdb.evaluate_detections(all_boxes, output_dir)
end = time.time()
print("test time: %0.4fs" % (end - start))