-
Notifications
You must be signed in to change notification settings - Fork 7
/
Copy pathdetect_frame.py
288 lines (233 loc) · 10.1 KB
/
detect_frame.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
# ====================================================
# @Author : Xiao Junbin
# @Email : [email protected]
# @File : detect_frame.py
# ====================================================
import os
import sys
sys.path.insert(0, 'lib')
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 matplotlib.pyplot as plt
# import torchvision
from collections import defaultdict
import os.path as osp
import pickle as pkl
# 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.utils.blob import im_list_to_blob
from model.faster_rcnn.resnet import resnet
import pdb
# import threading
class FeatureExtractor():
def __init__(self, train_loader, val_loader, cfg_file, classes,
class_agnostic, cuda, checkpoint_path, save_dir):
self.cfg_file = cfg_file
self.classes = classes
self.train_loader = train_loader
self.val_loader = val_loader
self.class_agnostic = class_agnostic
self.cuda = cuda
self.load_name = checkpoint_path
self.save_dir = save_dir
self.max_per_image = 100
self.pthresh = 0
def build_model(self):
self.fasterRCNN = resnet(self.classes, 101, pretrained=False, class_agnostic=self.class_agnostic)
self.fasterRCNN.create_architecture()
def load_checkpoint(self):
print('Load checkpoint from {}'.format(self.load_name))
if self.cuda > 0:
print('use gpu True')
checkpoint = torch.load(self.load_name)
else:
checkpoint = torch.load(self.load_name, map_location=(lambda storage, loc: storage))
self.fasterRCNN.load_state_dict(checkpoint['model'])
if 'pooling_mode' in checkpoint.keys():
cfg.POOLING_MODE = checkpoint['pooling_mode']
def run(self, mode):
cfg_from_file(self.cfg_file)
cfg.USE_GPU_NMS = self.cuda
# print('Using config:')
# pprint.pprint(cfg)
np.random.seed(cfg.RNG_SEED)
self.build_model()
self.load_checkpoint()
self.detect(mode)
def detect(self, mode):
im_data = torch.FloatTensor(1)
im_info = torch.FloatTensor(1)
num_boxes = torch.LongTensor(1)
gt_boxes = torch.FloatTensor(1)
# ship to cuda
if self.cuda > 0:
im_data = im_data.cuda()
im_info = im_info.cuda()
num_boxes = num_boxes.cuda()
gt_boxes = gt_boxes.cuda()
cfg.CUDA = True
self.fasterRCNN = self.fasterRCNN.cuda()
# make variable
with torch.no_grad():
im_data = Variable(im_data)
im_info = Variable(im_info)
num_boxes = Variable(num_boxes)
gt_boxes = Variable(gt_boxes)
if not osp.exists(self.save_dir):
os.makedirs(self.save_dir)
self.fasterRCNN.eval()
sample_loader = self.val_loader if mode == 'val' else self.train_loader
total_n = len(sample_loader)
print('Total number: {}'.format(total_n))
for iv, inputs in enumerate(sample_loader):
# if iv <= 200000: continue
# if iv > 200000: break
spatial_data, frame_name = inputs
frame_name = frame_name[0]
save_name = osp.join(self.save_dir, frame_name + '.pkl')
if osp.exists(save_name): continue
fdet = self.get_snippet_dets(spatial_data, im_data, im_info, gt_boxes, num_boxes)
dirname = osp.dirname(save_name)
if not osp.exists(dirname):
os.makedirs(dirname)
with open(save_name, 'wb') as fp:
pkl.dump(fdet, fp)
if iv % 500 == 0:
print('{}/{} {}'.format(iv, total_n, save_name))
def get_snippet_dets(self, spatial_data, im_data, im_info, gt_boxes, num_boxes):
"""
get detection results for each frame in the snippet
:param spatial_data:
:param im_data:
:param im_info:
:param gt_boxes:
:param num_boxes:
:return:
"""
fdet = {}
im_blob, im_scale = spatial_data['im_blob'][0], spatial_data['im_scale'][0]
im_info_np = np.array([[im_blob.shape[1], im_blob.shape[2], im_scale[0]]], dtype=np.float32)
im_data_pt = im_blob
im_data_pt = im_data_pt.permute(0, 3, 1, 2)
im_info_pt = torch.from_numpy(im_info_np)
im_data.data.resize_(im_data_pt.size()).copy_(im_data_pt)
im_info.data.resize_(im_info_pt.size()).copy_(im_info_pt)
gt_boxes.data.resize_(1, 1, 5).zero_()
num_boxes.data.resize_(1).zero_()
rois, cls_prob, bbox_pred, \
rpn_loss_cls, rpn_loss_box, \
RCNN_loss_cls, RCNN_loss_bbox, \
rois_label, pooled_feat, base_feat = self.fasterRCNN(im_data, im_info, gt_boxes, num_boxes)
scores = cls_prob.data
boxes = rois.data[:, :, 1:5]
pred_boxes = self.transform_bbox(boxes, bbox_pred, im_info, scores)
assert im_scale[0].item() != 0, "im_scale==0"
pred_boxes /= im_scale[0].item()
fdet['cls_prob'] = cls_prob.cpu().data.numpy()
fdet['bbox'] = pred_boxes.cpu().data.numpy()
fdet['roi_feat'] = pooled_feat.cpu().data.numpy()
# fdet['base_feat'] = base_feat.cpu().data.numpy()
return fdet
def transform_bbox(self, boxes, bbox_pred, im_info, scores):
"""
transform bbox from
:param boxes:
:param bbox_pred:
:param im_info:
:param scores:
:return:
"""
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 self.class_agnostic:
if self.cuda > 0:
box_deltas = box_deltas.view(-1, 4) * torch.FloatTensor(cfg.TRAIN.BBOX_NORMALIZE_STDS).cuda() \
+ torch.FloatTensor(cfg.TRAIN.BBOX_NORMALIZE_MEANS).cuda()
else:
box_deltas = box_deltas.view(-1, 4) * torch.FloatTensor(cfg.TRAIN.BBOX_NORMALIZE_STDS) \
+ torch.FloatTensor(cfg.TRAIN.BBOX_NORMALIZE_MEANS)
box_deltas = box_deltas.view(1, -1, 4)
else:
if self.cuda > 0:
box_deltas = box_deltas.view(-1, 4) * torch.FloatTensor(cfg.TRAIN.BBOX_NORMALIZE_STDS).cuda() \
+ torch.FloatTensor(cfg.TRAIN.BBOX_NORMALIZE_MEANS).cuda()
else:
box_deltas = box_deltas.view(-1, 4) * torch.FloatTensor(cfg.TRAIN.BBOX_NORMALIZE_STDS) \
+ torch.FloatTensor(cfg.TRAIN.BBOX_NORMALIZE_MEANS)
box_deltas = box_deltas.view(1, -1, 4 * 81)
# print(boxes.shape, box_deltas.shape)
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]))
return pred_boxes
def bbox_selection(self, relation, pred_boxes, scores, pooled_feat):
"""
delete bbox of low scores and do NMS
:param pred_boxes:
:param scores:
:return:
"""
scores = scores.squeeze()
pred_boxes = pred_boxes.squeeze()
class_bboxes = {}
class_feats = {}
class_pros = {}
sub, pre, obj = relation[0].split('-')
sind, oind = self.classes.index(sub), self.classes.index(obj)
for c, j in enumerate([sind, oind]):
inds = torch.nonzero(scores[:, j] > self.pthresh).view(-1)
if inds.numel() > 0:
cls_scores = scores[:, j][inds]
_, order = torch.sort(cls_scores, 0, True)
if self.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 = cls_dets[order]
keep = nms(cls_dets, cfg.TEST.NMS, force_cpu=not cfg.USE_GPU_NMS)
inds = keep.view(-1).long()
#inds = inds.numpy()
inds_new = []
for i in inds:
bbox = cls_dets[i, 0:4]
if bbox[2]-bbox[1] < 5.0 or bbox[3]-bbox[1] < 5.0:
continue
inds_new.append(i)
if len(inds_new) == 0: continue
inds = torch.cuda.LongTensor(np.array(inds_new))
cls_dets = cls_dets[inds]
cls_feats = pooled_feat[inds]
cls_pros = scores[inds]
class_bboxes[c] = cls_dets.data.cpu().numpy()
class_feats[c] = cls_feats.data.cpu().numpy()
class_pros[c] = cls_pros.data.cpu().numpy()
if c == 0 and sind == oind:
class_bboxes[1] = cls_dets.data.cpu().numpy()
class_feats[1] = cls_feats.data.cpu().numpy()
class_pros[1] = cls_pros.data.cpu().numpy()
break
"""
print('bbox shape:{}\t classme shape:{}\tfeat shape:{}'.format(cls_dets.shape,
cls_feats.shape,
cls_pros.shape))
"""
return class_bboxes, class_pros, class_feats