-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathtest.py
324 lines (302 loc) · 21.1 KB
/
test.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
import argparse
import json
import os
from itertools import permutations
import numpy as np
import torch
import torch.utils.data
import evaluation
from yolov3.datasets import MVCOCODataset, COCODatasetFromMV
from yolov3.epipolar_geometry import compute_fundamental_matrix
from yolov3.test import evaluate_singleview, evaluate_multiview
from yolov3.utils.parser import get_parser_from_arguments
from yolov3.yolo import YOLOv3
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
def test_multi_view(parser, f_matrices=None):
views = parser.views
perms = permutations(views, 2)
print("Runnning Test")
dataset = MVCOCODataset(parser.test["dir"],
views=views,
annotations_file=parser.test["annotation_file"],
multiscale=False,
normalized_labels=parser.test["normalized"],
img_size=parser.img_size,
padding_value=1)
if f_matrices is None:
coco = json.load(open(parser.train["annotation_file"], 'r'))
f_matrices = {tuple([*perm, c]): compute_fundamental_matrix(coco, *perm, img_size=parser.img_size, class_id=c)
for perm in perms for c in dataset.class_indices}
# Initiate model
model = YOLOv3(len(dataset.classes), anchors=parser.anchors).to(device)
if parser.test["weights_file"] is not None:
if parser.test["weights_file"].endswith(".weights"):
# Load darknet weights
model.load_darknet_weights(parser.test["weights_file"])
else:
# Load checkpoint weights
model.load_state_dict(torch.load(parser.test["weights_file"]))
model.eval()
_, precision, recall, AP, f1, ap_class, detections = evaluate_multiview(dataset,
model,
parser.test["iou_thres"],
parser.test["conf_thres"],
parser.test["nms_thres"],
parser.img_size,
parser.test["p_value"],
parser.workers,
f_matrices,
views,
bs=parser.test["batch_size"],
return_detections=True)
return detections
def test_single_view(parser):
print("Runnning Test")
views = parser.views
dataset = COCODatasetFromMV(parser.test["dir"],
annotations_file=parser.test["annotation_file"],
multiscale=False,
augment=False,
normalized_labels=parser.test["normalized"],
img_size=parser.img_size,
views=views,
padding_value=1)
# Initiate model
model = YOLOv3(len(dataset.classes), anchors=parser.anchors).to(device)
if parser.test["weights_file"] is not None:
if parser.test["weights_file"].endswith(".weights"):
# Load darknet weights
model.load_darknet_weights(parser.test["weights_file"])
else:
# Load checkpoint weights
model.load_state_dict(torch.load(parser.test["weights_file"]))
model.eval()
_, precision, recall, AP, f1, ap_class, detections = evaluate_singleview(
dataset,
model,
parser.test["iou_thres"],
parser.test["conf_thres"],
parser.test["nms_thres"],
parser.img_size,
parser.workers,
parser.test["weights_file"],
parser.test["batch_size"],
return_detections=True
)
return detections
def get_matrices():
"""
Fundamental matrices for ICPR 20 ms
:return: a dict of tuples, where the key is the tuple (src_view, dst_view, cls) and the values are the tuple
(fundamental matrix np.array, mu, sigma).
"""
return {('A', 'B', 1): (np.array([[-3.75529011e-08, -1.64293354e-07, -9.74453879e-03],
[-3.74756394e-07, 1.34064598e-09, 1.57336122e-04],
[9.93107299e-03, 3.36792723e-05, -6.13736306e-02]]), 0.03205558711190775,
1.8154441977086766),
('A', 'B', 2): (np.array([[3.30296215e-06, 9.14764774e-06, 1.74393296e-02],
[-6.15558318e-06, -8.88062077e-09, 1.79308157e-03],
[-2.03338030e-02, -2.69037439e-03, 5.67354058e-01]]), 0.11346449472459673,
2.0260713907656167),
('A', 'B', 3): (np.array([[3.52565778e-07, 6.96864463e-08, 1.01147058e-02],
[9.67420040e-07, 1.62526448e-08, -1.87062195e-04],
[-1.06183560e-02, -1.90382070e-04, 1.37796564e-01]]), -0.13677395228138325,
2.921226442890686),
('A', 'B', 4): (np.array([[-4.94489373e-07, 8.03437517e-07, -1.10325733e-02],
[-2.00188573e-06, 5.51333974e-09, 3.37829270e-04],
[1.16955226e-02, -1.67043254e-04, -1.05025273e-01]]), 0.12901145290341248,
3.2023637017144244),
('A', 'C', 1): (np.array([[5.26064024e-08, 5.70611245e-07, 1.01454353e-02],
[-2.25611831e-07, -9.33302786e-10, 6.25342258e-05],
[-1.02554035e-02, -1.99708109e-04, 4.09666266e-02]]), 0.275348956264891,
3.1666062577077474),
('A', 'C', 2): (np.array([[-9.07641204e-06, -3.99993373e-06, 2.42866321e-02],
[7.24449237e-06, -3.95906170e-08, -2.43765429e-03],
[-2.03082464e-02, 1.47741242e-03, -3.31080483e-01]]), 0.3480936328462325,
2.47390843273624),
('A', 'C', 3): (np.array([[3.57357616e-07, 4.84123336e-07, -1.10649227e-02],
[-1.26730910e-06, 1.09421595e-08, 4.30795627e-04],
[1.12670319e-02, -2.63971217e-04, -8.58912985e-02]]), -0.438201608570547,
4.160244995603259),
('A', 'C', 4): (np.array([[1.04760042e-06, -1.36787293e-06, 1.24688995e-02],
[-3.48711424e-06, 4.61248018e-09, 7.56825870e-04],
[-1.15425672e-02, 3.09475504e-04, -2.28394506e-01]]), -0.06972104123650755,
3.568695308089554),
('A', 'D', 1): (np.array([[3.27273286e-07, 2.20368492e-06, 1.01036740e-02],
[-3.23478204e-06, -1.28156788e-08, 9.87329136e-04],
[-1.00379860e-02, -7.07620661e-04, -3.03051879e-02]]), -0.521881975697132,
3.728220723889957),
('A', 'D', 2): (np.array([[1.28812955e-06, -6.92865718e-06, 2.42322504e-02],
[1.65189311e-06, -1.37223272e-07, -1.01614102e-04],
[-2.38118725e-02, 2.51191620e-03, -4.37838976e-01]]), -0.3967011185300554,
2.9853421354124885),
('A', 'D', 3): (np.array([[2.17233033e-06, -2.39818645e-06, -1.16049531e-02],
[4.86861969e-06, 1.73865356e-08, -1.67939656e-03],
[9.36391901e-03, 8.57716660e-04, 4.89118910e-01]]), 1.5639636078216945,
4.660567913821218),
('A', 'D', 4): (np.array([[-2.51122446e-06, -1.03560182e-05, 1.45297548e-02],
[9.66597899e-06, -3.29739677e-08, -3.79943830e-03],
[-1.25751451e-02, 4.12778390e-03, -3.94473340e-01]]), -1.1746109836991125,
2.8628894998687557),
('B', 'A', 1): (np.array([[2.64432625e-07, 7.75583153e-07, 9.62187119e-03],
[-3.64206366e-07, -9.60607376e-10, 7.99273750e-05],
[-9.88850299e-03, -1.95911083e-04, 4.85735622e-02]]), -0.31898143496560755,
1.8083630202494199),
('B', 'A', 2): (np.array([[7.17425914e-07, -1.44822503e-06, 2.09903332e-02],
[5.93184788e-06, 4.39579037e-08, -1.73245357e-03],
[-2.25413603e-02, 2.54604568e-04, 4.33710470e-01]]), -0.2050771043244269,
1.943396593969977),
('B', 'A', 3): (np.array([[-2.85666898e-07, 3.01813733e-07, 1.04511486e-02],
[5.73174385e-08, 4.06887307e-09, 1.13963014e-04],
[-1.04007717e-02, -1.34420667e-04, -1.91107865e-02]]), -0.27176686220405477,
2.903384017409568),
('B', 'A', 4): (np.array([[4.02725932e-06, 5.55994671e-06, 8.61881294e-03],
[-5.97232473e-06, -4.51744416e-08, 2.87917841e-03],
[-1.10660907e-02, -2.70446650e-03, 3.29638846e-01]]), 0.4104544268710088,
3.750871432034983),
('B', 'C', 1): (np.array([[9.75477889e-08, -4.47032114e-07, 1.07146613e-02],
[8.57049634e-07, 1.13526281e-09, -3.40172708e-04],
[-1.09574221e-02, 1.62832772e-04, 8.42631376e-02]]), -0.13373439100828727,
3.1924407161650605),
('B', 'C', 2): (np.array([[1.30114962e-06, -1.15348034e-05, 2.23272974e-02],
[-2.08378741e-06, 8.35108646e-08, 4.56970711e-04],
[-1.92935924e-02, 3.40211733e-03, -9.46003057e-01]]), 0.18155480815599456,
2.1868336531687698),
('B', 'C', 3): (np.array([[-1.63839192e-07, 1.26248265e-06, -1.10477182e-02],
[-1.02022471e-07, -1.32460169e-08, 1.57030454e-04],
[1.08911136e-02, -4.95729880e-04, 4.68922320e-02]]), -0.3483379604355908,
4.145804793810657),
('B', 'C', 4): (np.array([[4.09132751e-06, -2.87127507e-07, 1.15761681e-02],
[3.86178454e-06, -9.55595927e-08, -5.45396425e-04],
[-1.54124932e-02, 4.08106537e-04, 4.29291718e-01]]), -0.7344546992744537,
3.3840237272671945),
('B', 'D', 1): (np.array([[1.39140686e-07, -3.27879807e-06, -9.27326225e-03],
[2.60580928e-06, 3.61363997e-08, -6.58287448e-04],
[9.29988580e-03, 9.45744363e-04, -4.94251723e-02]]), 1.3172958063401152,
3.787840280905243),
('B', 'D', 2): (np.array([[-2.16472170e-05, 3.38850343e-05, 1.93633445e-02],
[-2.73892142e-05, 4.69305740e-08, 8.66398030e-03],
[-8.18328888e-03, -1.06627546e-02, -1.35927295e+00]]), -1.3022801110786415,
3.631925391556244),
('B', 'D', 3): (np.array([[7.68289444e-07, -3.01771722e-07, -1.12555369e-02],
[-1.97721265e-07, -6.84165708e-09, -4.43811592e-05],
[1.08448669e-02, 3.05964555e-04, 2.25465597e-02]]), 1.4399732854092697,
4.277485300518875),
('B', 'D', 4): (np.array([[6.59840905e-07, -1.58656655e-06, 1.12823519e-02],
[-8.68603658e-07, -7.26694714e-08, 6.58671296e-04],
[-1.09191852e-02, 2.71564673e-04, -2.25062645e-01]]), -0.5469439773016519,
2.966438449797102),
('C', 'A', 1): (np.array([[-2.21389885e-07, -3.39805454e-07, 1.08401460e-02],
[6.74995142e-07, 2.50850802e-09, -1.89645022e-04],
[-1.07190370e-02, 5.49897050e-05, -3.37057522e-03]]), -0.16258652307733729,
3.211435403983224),
('C', 'A', 2): (np.array([[3.55511919e-06, 2.37055964e-06, 1.80375321e-02],
[1.59115987e-05, -2.69443914e-07, -4.79702460e-03],
[-2.55998782e-02, -3.15618895e-04, 1.82865136e+00]]), -0.34762878190752555,
2.474094225867839),
('C', 'A', 3): (np.array([[-1.67638251e-07, -5.70725495e-07, -1.06034933e-02],
[-9.22070404e-07, 9.58186644e-09, 4.27845147e-04],
[1.11931219e-02, 1.51430486e-04, -1.97976311e-01]]), 0.14159620594749628,
4.118296999600756),
('C', 'A', 4): (np.array([[8.69087540e-07, -8.24349630e-08, 1.34237570e-02],
[3.36308589e-06, -1.29840012e-07, -9.87581683e-04],
[-1.49294731e-02, 5.91487648e-04, 1.58688279e-01]]), -0.017219615382833137,
3.5280186418925132),
('C', 'B', 1): (np.array([[5.74275849e-07, 2.27019599e-07, -1.05353390e-02],
[7.44412652e-08, -2.77061025e-10, -4.85988386e-05],
[1.00569906e-02, -1.88727560e-04, 1.42547004e-01]]), 0.07867992095006568,
3.0928105806584294),
('C', 'B', 2): (np.array([[5.25184901e-06, -4.20641661e-06, -2.00013053e-02],
[-7.03579769e-06, 8.81549320e-08, 2.25037229e-03],
[1.99176342e-02, 1.09573796e-03, -4.17869055e-01]]), 0.5281750571312332,
2.1723010444670163),
('C', 'B', 3): (np.array([[3.29348025e-07, 3.13290114e-07, -1.10559314e-02],
[4.54864082e-07, -4.23671631e-09, -5.99072501e-05],
[1.05502963e-02, -1.41546423e-04, 1.17026815e-01]]), 0.21682143382062896,
4.055402114768905),
('C', 'B', 4): (np.array([[3.19197441e-06, 5.09344158e-07, -1.39997945e-02],
[3.94077699e-07, -2.86996467e-09, -3.31136045e-05],
[1.16246692e-02, -1.36967414e-04, 3.71838200e-01]]), -1.0880981911735053,
3.2436267397779925),
('C', 'D', 1): (np.array([[4.38128592e-07, 1.06589517e-06, -1.10977000e-02],
[2.54116090e-07, -7.99504397e-09, -2.70987498e-05],
[1.03196788e-02, -4.55202180e-04, 2.35472040e-01]]), 1.8810482696880706,
5.311175580012618),
('C', 'D', 2): (np.array([[9.68795261e-06, 4.66548968e-06, -2.54563830e-02],
[-6.97979461e-06, 1.88206755e-07, 2.05455096e-03],
[1.92919129e-02, -2.01618600e-03, 1.18515415e+00]]), 0.9664900198985676,
3.0573652800729514),
('C', 'D', 3): (np.array([[-7.13047117e-07, 1.95157152e-06, 1.10473647e-02],
[1.57074906e-06, 4.41125985e-09, -5.05596705e-04],
[-1.17753301e-02, -6.58801911e-04, 3.25302617e-01]]), -1.5666239035894076,
6.141123361481922),
('C', 'D', 4): (np.array([[8.59017403e-07, 3.53872135e-06, -1.44078487e-02],
[3.83739283e-06, 1.47381710e-07, -1.66127522e-03],
[1.12858382e-02, -1.03292933e-03, 9.85144540e-01]]), 1.0940014594845542,
4.21160617479416),
('D', 'A', 1): (np.array([[-8.02539461e-08, -1.35836905e-06, 1.09014519e-02],
[2.34329317e-06, 3.97573931e-09, -8.29382265e-04],
[-1.11921256e-02, 4.61893510e-04, 1.11722859e-01]]), 0.8998493555858764,
3.784020238353342),
('D', 'A', 2): (np.array([[1.15935879e-05, 1.14203216e-05, -2.94456459e-02],
[-6.87737610e-06, -3.10031415e-07, 3.22827572e-03],
[2.05059905e-02, -4.15439144e-03, 1.56109421e+00]]), -0.5081076788384974,
3.13865911335172),
('D', 'A', 3): (np.array([[2.24817508e-07, -6.92959917e-06, -1.01244523e-02],
[1.91854199e-06, 9.88461938e-09, -5.42769473e-04],
[1.14949191e-02, 2.01428581e-03, -4.13982021e-01]]), -0.6627156595475379,
4.504198887599453),
('D', 'A', 4): (np.array([[-2.14811781e-07, -2.09570850e-06, -1.04368487e-02],
[3.30381113e-06, 3.99480859e-08, -1.19459427e-03],
[1.02585473e-02, 8.74510327e-04, 6.06082261e-02]]), -1.0039836970281002,
2.730519631832975),
('D', 'B', 1): (np.array([[-3.81091352e-07, -1.06578095e-07, -9.86106628e-03],
[-2.95171076e-07, 6.94422858e-09, 2.55585257e-04],
[1.02822418e-02, -1.41987551e-04, -9.08094064e-02]]), -1.6894215372413224,
3.557990627555916),
('D', 'B', 2): (np.array([[5.93913882e-06, -1.90086641e-05, 2.36204666e-02],
[2.46217259e-05, -1.40405459e-07, -8.06160889e-03],
[-2.90649540e-02, 6.43500822e-03, 1.06972129e+00]]), 0.3939760344450597,
2.8409123805751406),
('D', 'B', 3): (np.array([[5.45062905e-07, -1.36342319e-07, 1.07439771e-02],
[1.27555615e-06, 2.38676973e-08, -2.21960691e-04],
[-1.13307389e-02, -1.87130991e-04, 1.31050900e-01]]), 0.43557786319948477,
4.525729597548589),
('D', 'B', 4): (np.array([[-1.19898465e-06, 1.62799195e-05, -1.48893878e-02],
[-1.84420908e-05, 2.01989698e-07, 7.87537477e-03],
[1.62367513e-02, -7.18620917e-03, -2.98094954e-01]]), -0.7088220967150424,
3.7208145973646887),
('D', 'C', 1): (np.array([[6.64150921e-08, 2.72874992e-07, -1.03066494e-02],
[-4.16126389e-07, -9.87432753e-10, 1.09595503e-04],
[1.03931191e-02, -4.67774611e-05, -4.19556131e-02]]), -1.9184124755111516,
5.219067831447366),
('D', 'C', 2): (np.array([[3.72171867e-06, 1.26576084e-05, -2.50727901e-02],
[-3.47590415e-06, -3.86459518e-07, 1.91361135e-03],
[2.01930255e-02, -4.18886605e-03, 9.81755972e-01]]), -0.36202404725895915,
3.3659456137724417),
('D', 'C', 3): (np.array([[5.99663737e-07, -2.93921394e-06, 1.31779892e-02],
[-6.96507916e-06, 2.84278669e-08, 2.04466636e-03],
[-1.02763011e-02, 9.08195055e-04, -9.22304508e-01]]), 1.1294191584727113,
5.960453878007358),
('D', 'C', 4): (np.array([[8.33792353e-06, 5.80033116e-06, 8.69666269e-03],
[1.39940590e-05, -1.03665120e-08, -5.32594000e-03],
[-2.16891569e-02, -2.20100976e-03, 3.72232429e+00]]), 1.254158912335075,
6.58554087255596)}
if __name__ == '__main__':
args = argparse.ArgumentParser()
args.add_argument("--gt_path", help="YAML Config file path")
args.add_argument("--type", choices=["sv", "mv"],
help="Testing type: mv (default_ for multi-view and sv for single-view", default="mv")
p, opt = get_parser_from_arguments(args)
gt_path = opt.gt_path
if opt.type == "sv":
preds = test_single_view(p)
elif opt.type == "mv":
preds = test_multi_view(p, f_matrices=get_matrices())
else:
raise ValueError(f"Invalid mode {opt.type}")
gt_path = opt.gt_path
if gt_path:
evaluation.evaluate(gt_path, preds)
json_file_name = os.path.join(p.db_name, p.test["json_file_output"])
json.dump(preds, open(json_file_name, 'w'))