Sort.cpp
19.2 KB
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
#include "Sort.h"
#include <opencv2/opencv.hpp>
//#include "ImageSaveGPU.h"
#ifdef _MSC_VER
#include <io.h>
#include <direct.h>
#define _ACCESS _access
#define _MKDIR(a) _mkdir((a))
#else
#include <unistd.h>
#include <stdarg.h>
#include <sys/stat.h>
#define _ACCESS access
#define _MKDIR(a) mkdir((a),0755)
#endif
#include <time.h>
const int color[11][3] = { { 255, 0, 0 }, { 255, 128, 255 }, { 255, 128, 0 }, { 255, 215, 0 }, { 154, 205, 50 }, { 0, 128, 0 }, \
{0, 128, 255}, { 186, 85, 211 }, { 91, 46, 0 }, { 0, 0, 0 }, { 255, 255, 255 } };
Sort::Sort()
{
max_age = 10;
min_hits = 3;
frame_count = 0;
WORK = true;
//-----------------by zl---------------------//
istraffic = false; //by zl 不统计交通量
trackcount = 0;
//-----------------by zl---------------------//
max_track_length = MAX_LENGTH;
}
int Sort::update(int width, int height, bool isUseDet, vector< vector<float> > &dets, VPT_ObjInfo *result, vector<int> &deleteObjectID)
{
//cout << "track size: " << trackers.size() << endl;
//get predicted locations from existing trackers.
vector< vector<float> > trks;
vector<float> pos;
int ObjCount = 0; //by zl 本帧图像中的有效前景个数
vector<float> bbox;
for (int i = 0; i < trackers.size(); i++ )
{
pos = trackers[i].predict();
pos.push_back(1);
pos.push_back(trackers[i].cls);
trks.push_back(pos);
pos.clear();
}
if (isUseDet == true)
{
vector< vector<int> > matched;
vector<int> unmatched_dets;
vector<int> unmatched_trks;
//dets, x1,y1,x2,y2,score,cls
//trks, x1,y1,x2,y2,score,cls
Sort::associate_detections_to_trackers(matched, unmatched_dets, unmatched_trks, dets, trks, 0.3);
//std::cout<<__FILE__<<" matched:"<<matched.size()<<"\n";
//std::cout<<__FILE__<<" unmatched:"<<unmatched_dets.size()<<"\n";
//update matched trackers with assigned detections
for (int matched_number = 0; matched_number < matched.size(); matched_number++)
{
trackers[matched[matched_number][1]].update(dets[matched[matched_number][0]]);
trackers[matched[matched_number][1]].score = dets[matched[matched_number][0]][4];
trackers[matched[matched_number][1]].cls = dets[matched[matched_number][0]][5];
//cout << dets[matched[matched_number][0]][4] << endl;
//trackers[matched[matched_number][1]].update(trks[matched[matched_number][1]]);
}
//create and initialise new trackers for unmatched detections
for (int unmatched_dets_number = 0; unmatched_dets_number < unmatched_dets.size(); unmatched_dets_number++)
{
KalmanBoxTracker tracker = KalmanBoxTracker(dets[unmatched_dets[unmatched_dets_number]], max_track_length);
tracker.id = -1;
/*tracker.id = trackcount;
trackcount += 1;*/
trackers.push_back(tracker);
trackers[trackers.size() - 1].score = dets[unmatched_dets[unmatched_dets_number]][4];//by zl 20170525 解决第一次检测时置信度为0问题
// cout << "trackers size: " << trackers.size() << endl;
}
for (int trackers_number = 0; trackers_number < trackers.size();)
{
//cout << trackers[trackers_number].id << " " <<trackers[trackers_number].time_since_update << " " << trackers[trackers_number].hit_streak << " " << min_hits << " " << frame_count << endl;
if (trackers[trackers_number].time_since_update > max_age)
{
if (trackers[trackers_number].id != -1)
{
//cout << "----------------------------- delete id: " << trackers[trackers_number].id << endl;
deleteObjectID.push_back(trackers[trackers_number].id);
}
trackcount++;
trackers.erase(trackers.begin() + trackers_number);
continue;
}
if ((trackers[trackers_number].time_since_update < FusionInterval) && ((trackers[trackers_number].hit_streak >= min_hits) || (frame_count <= (min_hits*FusionInterval))))
{
if (trackers[trackers_number].id == -1)
trackers[trackers_number].id = trackcount++;
result[ObjCount].id = trackers[trackers_number].id;
bbox = trackers[trackers_number].get_state();
result[ObjCount].left = bbox[0]; // bbout[i][0];
result[ObjCount].top = bbox[1]; //bbout[i][1];
result[ObjCount].right = bbox[2]; //bbout[i][2];
result[ObjCount].bottom = bbox[3]; //bbout[i][3];
result[ObjCount].confidence = trackers[trackers_number].score; // bbout[i][4];
result[ObjCount].index = trackers[trackers_number].cls - 1; // bbout[i][5] - 1;
RectboundCheck(width, height, &result[ObjCount]);
result[ObjCount].center_x = result[ObjCount].left + (result[ObjCount].right - result[ObjCount].left) * 0.5; // 中心点 add by 20170227
result[ObjCount].center_y = result[ObjCount].top + (result[ObjCount].bottom - result[ObjCount].top) * 0.5; // 中心点
if(trackers[trackers_number].age==2*FusionInterval)
{
result[ObjCount].snap_flag = 1; // 中心点
}
else
{
result[ObjCount].snap_flag = 0; // 中心点
}
#if _Debug
printf("trackers_number = %d, trackers.size() = %d, update: index = %d, id = %d, (%d, %d), (%d, %d)\n", trackers_number, trackers.size(), result[ObjCount].index, result[ObjCount].id, result[ObjCount].left, result[ObjCount].top, result[ObjCount].right, result[ObjCount].bottom);
#endif
ObjCount++;
}
trackers_number++;//共多少条轨迹
}
}
else
{
for (int trackers_number = 0; trackers_number < trackers.size(); trackers_number++)
{
if (trackers[trackers_number].id == -1)
trackers[trackers_number].id = trackcount++;
bbox = trackers[trackers_number].get_state();
result[trackers_number].id = trackers[trackers_number].id;
result[trackers_number].left = bbox[0]; // bbout[i][0];
result[trackers_number].top = bbox[1]; //bbout[i][1];
result[trackers_number].right = bbox[2]; //bbout[i][2];
result[trackers_number].bottom = bbox[3]; //bbout[i][3];
result[trackers_number].confidence = trackers[trackers_number].score; // bbout[i][4];
result[trackers_number].index = bbox[5] - 1;// trackers[trackers_number].cls - 1; // bbout[i][5] - 1;
RectboundCheck(width, height, &result[trackers_number]);
result[trackers_number].center_x = (int)(result[trackers_number].left + (result[trackers_number].right - result[trackers_number].left) * 0.5); // 中心点 add by 20170227
result[trackers_number].center_y = (int)(result[trackers_number].top + (result[trackers_number].bottom - result[trackers_number].top) * 0.5); // 中心点
result[ObjCount].snap_flag = 0; // 中心点
ObjCount++;
}
}
//---------------------------注释掉了这步操作 用了新的绘制轨迹的函数 需要绘制调用addTracker(Mat *img)方法 by lm---------------------------------------------/
//addTracker(result, ObjCount);
frame_count += 1; //帧数加一
return ObjCount;
}
//
////---------------------------by zl ---------------------------------------------/
//int Sort::addTracker(VPT_ObjInfo *result, int resultcount)
//{
// for (int i = 0; i < resultcount; i++)
// {
// bool flag = false;
// for (int j = 0; j < tracker.size(); j++)
// {
// if (result[i].id == tracker[j].id) //若有匹配的 则更新
// {
// tracker[j].listinfo.push_back(result[i]);
// tracker[j].lost = 0;
// tracker[j].isupdate = true;
// tracker[j].num++; //每个ID的计数
// result[i].num = tracker[j].num;
//
// if (tracker[j].istraffic == false && istraffic)
// {
// int listmax = tracker[j].listinfo.size() - 1;
// }
// flag = true;
// break;
// }
// }
// if (!flag) //没有找到匹配项 新添加
// {
// mylist m_list;
// m_list.listinfo.push_back(result[i]);
// m_list.isupdate = true;
// m_list.lost = 0;
// m_list.id = result[i].id;
// m_list.index = result[i].index;;
// m_list.istraffic = false;
// m_list.num = 0; //20170306
// result[i].num = 0;
// m_list.startframe = frame_count;
//#if _Debug
// printf("addTracker_pushback :index = %d, id = %d, (%d, %d), (%d, %d)\n", m_list.index, m_list.id, result[i].left, result[i].top, result[i].right, result[i].bottom);
//#endif
// tracker.push_back(m_list);
// }
// }
// //绘制轨迹部分
//
// for (vector <mylist>::iterator iter = tracker.begin(); iter != tracker.end();)
// {
// if (iter->isupdate == false) //未更新的不绘制轨迹
// {
// iter->lost++;
// if (iter->lost > LOSTMAXFRAMECCOUNT) //若丢失太多 则删除该轨迹
// {
// iter->endframe = frame_count; //结束帧
// iter = tracker.erase(iter);//删除
// continue;
// }
// }
// else //只对更新后的绘制轨迹
// {
// ;
// }
// iter->isupdate = false;
// iter++;
// }
//
// return 1;
//}
//---------------------------利用trackers中的history 绘制路径 -by lm ---------------------------------------------/
//固定长度的轨迹，采用循环队列，仅保存目前最前N length的轨迹，避免对于停留在画面中目标 导致的内存一直增长
int Sort::addTracker(cv::Mat *img)
{
map<int, pair<int, int>> tracker;
vector<float> bbox;
int x_1, y_1, x_2, y_2;
for (auto iter : trackers)
{
if (iter.time_since_update < FusionInterval)
{
int index = iter.history.getFront();
int trackerSize = iter.history.size();
for (int i = 0; i < trackerSize; i++)
{
if (i == 0)
{
x_1 = iter.history.get(index).x;
y_1 = iter.history.get(index).y;
}
else
{
x_2 = iter.history.get(index).x;
y_2 = iter.history.get(index).y;
int colorIndex = iter.id % 11;
cv::line(*img, cvPoint(x_1, y_1), cvPoint(x_2, y_2), cvScalar(color[colorIndex][0], color[colorIndex][1], color[colorIndex][2]), 1);
//drawLineOnGPU()
x_1 = x_2;
y_1 = y_2;
}
//if (i == 0)
//{
// x_1 = iter.history.get(index)[0] + (iter.history.get(index)[2] - iter.history.get(index)[0]) / 2;
// y_1 = iter.history.get(index)[1] + (iter.history.get(index)[3] - iter.history.get(index)[1]) / 2;
//}
//else
//{
// x_2 = iter.history.get(index)[0] + (iter.history.get(index)[2] - iter.history.get(index)[0]) / 2;
// y_2 = iter.history.get(index)[1] + (iter.history.get(index)[3] - iter.history.get(index)[1]) / 2;
// int colorIndex = iter.id % 11;
// cv::line(*img, cvPoint(x_1, y_1), cvPoint(x_2, y_2), cvScalar(color[colorIndex][0], color[colorIndex][1], color[colorIndex][2]), 1);
// //drawLineOnGPU()
// x_1 = x_2;
// y_1 = y_2;
//}
index = (index + 1) % trackerSize;
}
//int index = iter.history->getFront();
//int trackerSize = iter.history->size();
//for (int i = 0; i < trackerSize; i++)
//{
// if (i == 0)
// {
// x_1 = iter.history->get(index)[0] + (iter.history->get(index)[2] - iter.history->get(index)[0]) / 2;
// y_1 = iter.history->get(index)[1] + (iter.history->get(index)[3] - iter.history->get(index)[1]) / 2;
// }
// else
// {
// x_2 = iter.history->get(index)[0] + (iter.history->get(index)[2] - iter.history->get(index)[0]) / 2;
// y_2 = iter.history->get(index)[1] + (iter.history->get(index)[3] - iter.history->get(index)[1]) / 2;
// int colorIndex = iter.id % 11;
// cv::line(*img, cvPoint(x_1, y_1), cvPoint(x_2, y_2), cvScalar(color[colorIndex][0], color[colorIndex][1], color[colorIndex][2]), 1);
// //drawLineOnGPU()
// x_1 = x_2;
// y_1 = y_2;
// }
// index = (index + 1) % trackerSize;
//}
}
}
return 1;
}
//不固定长度的轨迹版本 采用vector，轨迹一直保留，对于停留在画面中的物体，会有内存一直增长的隐患
//int Sort::addTracker(cv::Mat *img)
//{
// map<int, pair<int, int>> tracker;
// vector<float> bbox;
//
//
// int x_1, y_1, x_2, y_2;
// for (auto iter : trackers)
// {
// if (iter.time_since_update < FusionInterval)
// {
// for (int i = 0; i < iter.history.size(); i++)
// {
//
// if (i == 0)
// {
// x_1 = iter.history[i][0] + (iter.history[i][2] - iter.history[i][0]) / 2;
// y_1 = iter.history[i][1] + (iter.history[i][3] - iter.history[i][1]) / 2;
// }
// else
// {
// x_2 = iter.history[i][0] + (iter.history[i][2] - iter.history[i][0]) / 2;
// y_2 = iter.history[i][1] + (iter.history[i][3] - iter.history[i][1]) / 2;
// int colorIndex = iter.id % 11;
// cv::line(*img, cvPoint(x_1, y_1), cvPoint(x_2, y_2), cvScalar(color[colorIndex][0], color[colorIndex][1], color[colorIndex][2]), 1);
//
// //drawLineOnGPU()
//
// x_1 = x_2;
// y_1 = y_2;
// }
// }
// }
//
// }
//
// return 1;
//}
void Sort::Release()
{
//tracker.clear();
//vector <mylist>().swap(tracker);
trackers.clear();
vector<KalmanBoxTracker>().swap(trackers);
}
bool Sort::GetState()
{
return WORK;
}
void Sort::Pause()
{
WORK = false;
}
void Sort::ReSet()
{
WORK = true;
//Release();
//max_age = 1;
//min_hits = 3;
//frame_count = 0;
////-----------------by zl---------------------//
//istraffic = false; //by zl 不统计交通量
//trackcount = 0;
////-----------------by zl---------------------//
}
//---------------------------by zl ---------------------------------------------/
bool line_rect_intersection(cv::Point start_p, cv::Point end_p, int left, int top, int right, int bottom)
{
int a = start_p.y - end_p.y;
int b = end_p.x - start_p.x;
int c = start_p.x* end_p.y - end_p.x* start_p.y;
////思路:先看线段所在直线是否与矩形相交，如果不相交则必为 “Ｆ”，
////如果相交，则看线段的两个点是否在矩形的同一边（即两点的 x（y) 坐标都比矩形的小 x(y) 坐标小，或者大）,
////若在同一边则为“Ｆ”，否则就是相交的情况。
if ((a* left + b*top + c >= 0 && a* right + b* bottom + c <= 0) ||
(a* left + b*top + c <= 0 && a* right + b* bottom + c >= 0) ||
(a* left + b*bottom + c >= 0 && a* right + b* top + c <= 0) ||
(a* left + b*bottom + c >= 0 && a* right + b* top + c <= 0))
{
if (left > right)
{
swap(left, right);
}
if (top < bottom)
{
swap(top, bottom);
}
if ((start_p.x < left && end_p.x < left) ||
(start_p.x > right && end_p.x < left) ||
(start_p.y > top && end_p.y > top) ||
(start_p.y < bottom && end_p.y < bottom)) ///判断线段是否在矩形一侧
{
return false;
}
else
{
return true;
}
}
else
{
return false;
}
}
void RectboundCheck(int Width, int Height, VPT_ObjInfo * result) //防止坐标越界 by zl
{
if (result->left < 0)
result->left = 0;
if (result->left >= Width)
result->left = Width;
if (result->top < 0)
result->top = 0;
if (result->top >= Height)
result->top = Height;
if (result->right <= result->left)
result->right = result->left + 1;
if (result->right >= Width)
result->right = Width;
if (result->bottom < result->top)
result->bottom = result->top + 1;
if (result->bottom >= Height)
result->bottom = Height;
}
//------------------------------------其他函数----------------------------------------//
void Sort::associate_detections_to_trackers(vector< vector<int> > &matched, vector<int> &unmatched_dets, vector<int> &unmatched_trks, vector< vector<float> > &dets, vector< vector<float> > &trks, float iou_threshold)
{
if (0 == trks.size())
{
for (int x = 0; x < dets.size(); x++)
{
unmatched_dets.push_back(x);
}
}
else if (0 == dets.size())
{
for (int x = 0; x < trks.size(); x++)
{
unmatched_trks.push_back(x);
}
}
else
{
cv::Mat IoUMat(dets.size(), trks.size(), CV_32FC1);
for (int i = 0; i < dets.size(); i++)
for (int j = 0; j < trks.size(); j++)
{
//cls区分
if (1)
//if (dets[i][5] == trks[j][5])
{
IoUMat.at<float>(i, j) = IoU(dets[i], trks[j]);
}
else
{
IoUMat.at<float>(i, j) = 0;
}
}
//匈牙利算法
vector<int> assignment;
munkres(IoUMat, assignment);
vector<int>::iterator iter;
for (int trackers_indices = 0; trackers_indices < trks.size(); trackers_indices++)
{
iter = find(assignment.begin(), assignment.end(), trackers_indices);
if (iter == assignment.end())
{
//assignment中不存在trackers_indices值
unmatched_trks.push_back(trackers_indices);
}
}
vector<int> matched_row_col;
for (int detections_indices = 0; detections_indices < assignment.size(); detections_indices++)
{
if (assignment[detections_indices] == -1)
{
unmatched_dets.push_back(detections_indices);
}
else if (IoUMat.at<float>(detections_indices, assignment[detections_indices]) > iou_threshold)
{
matched_row_col.push_back(detections_indices);
matched_row_col.push_back(assignment[detections_indices]);
matched.push_back(matched_row_col);
matched_row_col.clear();
}
else
{
unmatched_dets.push_back(detections_indices);
unmatched_trks.push_back(assignment[detections_indices]);
}
}
}
}
//判断两条线是否相交
///------------alg 2------------
//叉积
double mult(cv::Point a, cv::Point b, cv::Point c)
{
return (a.x - c.x)*(b.y - c.y) - (b.x - c.x)*(a.y - c.y);
}
//aa, bb为一条线段两端点 cc, dd为另一条线段的两端点 相交返回true, 不相交返回false
bool intersect(cv::Point aa, cv::Point bb, cv::Point cc, cv::Point dd)
{
if (max(aa.x, bb.x)<min(cc.x, dd.x))
{
return false;
}
if (max(aa.y, bb.y)<min(cc.y, dd.y))
{
return false;
}
if (max(cc.x, dd.x)<min(aa.x, bb.x))
{
return false;
}
if (max(cc.y, dd.y)<min(aa.y, bb.y))
{
return false;
}
if (mult(cc, bb, aa)*mult(bb, dd, aa)<0)
{
return false;
}
if (mult(aa, dd, cc)*mult(dd, bb, cc)<0)
{
return false;
}
return true;
}
///------------alg 2------------