Hu Chunming / village_ascend_arm

Commit 581a68a4b0aa1e76c84194edf2fdcd61b0d49766

Authored by Hu Chunming
1 parent 8d319775

修正parse_road无返回值导致崩溃的问题； 

返回值中sy_point更换为cv::Point；
优化model_path
Inline Side-by-side
Showing 21 changed files with 114 additions and 1027 deletions
jni/VehicleNativeInterface.cpp
  View file @581a68a
... ... @@ -126,28 +126,22 @@ JNIEXPORT jint JNICALL Java_com_objecteye_nativeinterface_vehicle_VehicleNativeI
126 126 //�㷨���ò���
127 127 jclass cls_vehcileAnalysisParam = env->GetObjectClass(vehicleAnalysisParam);
128 128  
129   - jfieldID fid_dbpath = env->GetFieldID(cls_vehcileAnalysisParam, "dbPath", "Ljava/lang/String;");
130   - jfieldID fid_modelspath = env->GetFieldID(cls_vehcileAnalysisParam, "modelsPath", "Ljava/lang/String;");
  129 + jfieldID fid_sdk_path = env->GetFieldID(cls_vehcileAnalysisParam, "sdk_path", "Ljava/lang/String;");
131 130 jfieldID fid_gpuid = env->GetFieldID(cls_vehcileAnalysisParam, "gpuId", "I");
132 131  
133   - jstring str_dbpath = (jstring)env->GetObjectField(vehicleAnalysisParam, fid_dbpath);
134   - const char *dbpath = env->GetStringUTFChars(str_dbpath, JNI_FALSE);
135   - jstring str_modlespath = (jstring)env->GetObjectField(vehicleAnalysisParam, fid_modelspath);
136   - const char *models_path = env->GetStringUTFChars(str_modlespath, JNI_FALSE);
  132 + jstring str_sdk_path = (jstring)env->GetObjectField(vehicleAnalysisParam, fid_sdk_path);
  133 + const char *sdk_path = env->GetStringUTFChars(str_sdk_path, JNI_FALSE);
137 134 jint gpuid = env->GetIntField(vehicleAnalysisParam, fid_gpuid);
138 135  
139 136 void *vaHandle = NULL;
140 137 VillageParam param;
141 138 param.dev_id = gpuid;
142   - param.db_path = dbpath;
143   - param.model_path = models_path;
  139 + param.sdk_path = sdk_path;
144 140 int ret = village_pic_init(&vaHandle, param);
145   - if (ret != SUCCESS)
146   - {
  141 + if (ret != SUCCESS) {
147 142 printf("jni info:va_init failed.");
148 143 return ret;
149 144 }
150   -
151 145  
152 146 if (ret == SUCCESS)
153 147 {
... ... @@ -166,8 +160,7 @@ JNIEXPORT jint JNICALL Java_com_objecteye_nativeinterface_vehicle_VehicleNativeI
166 160 temp[0] = (jlong)vaHandle;
167 161 env->SetLongArrayRegion(handle, 0, 1, temp);
168 162 }
169   - env->ReleaseStringUTFChars(str_dbpath, dbpath);
170   - env->ReleaseStringUTFChars(str_modlespath, models_path);
  163 + env->ReleaseStringUTFChars(str_sdk_path, sdk_path);
171 164 return ret;
172 165 }
173 166  
... ... @@ -564,7 +557,7 @@ JNIEXPORT jint JNICALL Java_com_objecteye_nativeinterface_vehicle_VehicleNativeI
564 557 for (int k = 0; k < line_info.vec_pt.size(); k++)
565 558 {
566 559 auto pt = line_info.vec_pt[k];
567   - jobject line_info_point = env->NewObject(cls_SyPoint, mid_SyPoint, pt.x_, pt.y_);
  560 + jobject line_info_point = env->NewObject(cls_SyPoint, mid_SyPoint, pt.x, pt.y);
568 561 env->SetObjectArrayElement(ptArray, k, line_info_point);
569 562 }
570 563  
... ... @@ -584,7 +577,7 @@ JNIEXPORT jint JNICALL Java_com_objecteye_nativeinterface_vehicle_VehicleNativeI
584 577 for (int k = 0; k < road_info.vec_pt.size(); k++)
585 578 {
586 579 auto pt = road_info.vec_pt[k];
587   - jobject road_info_point = env->NewObject(cls_SyPoint, mid_SyPoint, pt.x_, pt.y_);
  580 + jobject road_info_point = env->NewObject(cls_SyPoint, mid_SyPoint, pt.x, pt.y);
588 581 env->SetObjectArrayElement(ptArray, k, road_info_point);
589 582 }
590 583  
... ...
src/PicAnalysis.cpp
  View file @581a68a
... ... @@ -19,7 +19,7 @@ int PicAnalysis::init(VillageParam param) {
19 19  
20 20 int ret = SY_FAILED;
21 21  
22   - ret = m_vehicle_analysis.init(dev_id, 16);
  22 + ret = m_vehicle_analysis.init(dev_id, param.sdk_path, 16);
23 23 if(0 != ret){
24 24 return -1;
25 25 }
... ... @@ -27,37 +27,38 @@ int PicAnalysis::init(VillageParam param) {
27 27 // head_tail_param ht_param;
28 28 // ht_param.devId = dev_id;
29 29 // ht_param.max_batch = 16;
  30 + // ht_param.sdk_path = param.sdk_path;
30 31 // ret = m_head_tail_algorithm.init(ht_param);
31 32 // if(0 != ret){
32 33 // return -1;
33 34 // }
34 35  
35   - ret = m_clothes_algorithm.init(dev_id);
  36 + ret = m_clothes_algorithm.init(dev_id, param.sdk_path);
36 37 if(0 != ret){
37 38 return -1;
38 39 }
39 40  
40   - ret = m_human_algorithm.init(dev_id);
  41 + ret = m_human_algorithm.init(dev_id, param.sdk_path);
41 42 if(0 != ret){
42 43 return -1;
43 44 }
44 45  
45   - ret = m_human_car_algorithm.init(dev_id);
  46 + ret = m_human_car_algorithm.init(dev_id, param.sdk_path);
46 47 if(0 != ret){
47 48 return -1;
48 49 }
49 50  
50   - ret = m_motor_rainshed_algorithm.init(dev_id);
  51 + ret = m_motor_rainshed_algorithm.init(dev_id, param.sdk_path);
51 52 if(0 != ret){
52 53 return -1;
53 54 }
54 55  
55   - ret = m_motor_phone_algorithm.init(dev_id);
  56 + ret = m_motor_phone_algorithm.init(dev_id, param.sdk_path);
56 57 if(0 != ret){
57 58 return -1;
58 59 }
59 60  
60   - ret = m_road_seg_algorithm.init(dev_id);
  61 + ret = m_road_seg_algorithm.init(dev_id, param.sdk_path);
61 62 if(0 != ret){
62 63 return -1;
63 64 }
... ... @@ -356,9 +357,9 @@ vector&lt;AnalysisResult&gt; PicAnalysis::analysis_img(vector&lt;sy_img&gt; vec_img){
356 357 }
357 358  
358 359 //压黄实线
359   - result_info.cross_line = m_road_seg_algorithm.check_cross_line(road_info.vec_line, vehicle_rect);
  360 + result_info.cross_line = m_road_seg_algorithm.check_cross_line(road_info.vec_line, vehicle_rect, src.width, src.height);
360 361 // 压导流线
361   - result_info.cross_diversion_line = m_road_seg_algorithm.check_cross_region(road_info.vec_road, vehicle_rect, 3); //3是导流线区域
  362 + result_info.cross_diversion_line = m_road_seg_algorithm.check_cross_region(road_info.vec_road, vehicle_rect, src.width, src.height, 3); //3是导流线区域
362 363 }
363 364 }
364 365  
... ...
src/ai_engine_module/HumanAnalysis.cpp
  View file @581a68a
... ... @@ -11,12 +11,14 @@ HumanAnalysis::~HumanAnalysis()
11 11 release();
12 12 }
13 13  
14   -int HumanAnalysis::init(int devId){
  14 +int HumanAnalysis::init(int devId, std::string sdk_root){
15 15 ACL_CALL(aclrtCreateContext(&ctx, devId), SY_SUCCESS, SY_FAILED);
16 16 ACL_CALL(aclrtSetCurrentContext(ctx), SY_SUCCESS, SY_FAILED);
17 17  
  18 + std::string model_path = sdk_root + "/models/hp/hp220908_310p.om";
  19 +
18 20 hp_param param;
19   - param.modelNames = "./models/hp/hp220908_310p.om";
  21 + param.modelNames = (char*)model_path.data();
20 22 param.devId = devId;
21 23  
22 24 cout << "hp_init start " << endl;
... ...
src/ai_engine_module/HumanAnalysis.h
  View file @581a68a
... ... @@ -14,7 +14,7 @@ public:
14 14 HumanAnalysis(/* args */);
15 15 ~HumanAnalysis();
16 16  
17   - int init(int devId);
  17 + int init(int devId, std::string sdk_root);
18 18  
19 19 vector<BodyColorInfo> detect(vector<sy_img> vec_img);
20 20  
... ...
src/ai_engine_module/HumanCarAnalysis.cpp
  View file @581a68a
... ... @@ -13,12 +13,14 @@ HumanCarAnalysis::~HumanCarAnalysis()
13 13 release();
14 14 }
15 15  
16   -int HumanCarAnalysis::init(int devId){
  16 +int HumanCarAnalysis::init(int devId, std::string sdk_root){
17 17 ACL_CALL(aclrtCreateContext(&ctx, devId), SY_SUCCESS, SY_FAILED);
18 18 ACL_CALL(aclrtSetCurrentContext(ctx), SY_SUCCESS, SY_FAILED);
19 19  
  20 + std::string model_path = sdk_root + "/models/hcp/hcp211008_310p.om";
  21 +
20 22 hcp_param param;
21   - param.modelNames = "./models/hcp/hcp211008_310p.om";
  23 + param.modelNames = (char*)model_path.data();
22 24 param.devId = devId;
23 25  
24 26 cout << "hcp_init start " << endl;
... ...
src/ai_engine_module/HumanCarAnalysis.h
  View file @581a68a
... ... @@ -41,7 +41,7 @@ public:
41 41 HumanCarAnalysis(/* args */);
42 42 ~HumanCarAnalysis();
43 43  
44   - int init(int devId);
  44 + int init(int devId, std::string sdk_root);
45 45  
46 46 std::vector<HumanCarResult> detect(vector<sy_img> vec_img);
47 47  
... ...
src/ai_engine_module/MotorPhoneAnalysis.cpp
  View file @581a68a
... ... @@ -10,11 +10,13 @@ MotorPhoneAnalysis::~MotorPhoneAnalysis()
10 10 release();
11 11 }
12 12  
13   -int MotorPhoneAnalysis::init(int devId){
  13 +int MotorPhoneAnalysis::init(int devId, std::string sdk_root){
14 14 ACL_CALL(aclrtCreateContext(&ctx, devId), SY_SUCCESS, SY_FAILED);
15 15  
  16 + std::string model_path = sdk_root + "/models/motor_phone/motor_phone1127_310p.om";
  17 +
16 18 motor_phone_param param;
17   - param.modelNames = "./models/motor_phone/motor_phone1127_310p.om";
  19 + param.modelNames = (char*)model_path.data();
18 20 param.thresld = 0.25;
19 21 param.devId = devId;
20 22  
... ...
src/ai_engine_module/MotorPhoneAnalysis.h
  View file @581a68a
... ... @@ -11,7 +11,7 @@ public:
11 11 MotorPhoneAnalysis(/* args */);
12 12 ~MotorPhoneAnalysis();
13 13  
14   - int init(int devId);
  14 + int init(int devId, std::string sdk_root);
15 15  
16 16 vector<MotorPhoneResult> detect(vector<sy_img> vec_img);
17 17  
... ...
src/ai_engine_module/MotorRainshedAnalysis.cpp
  View file @581a68a
... ... @@ -10,11 +10,13 @@ MotorRainshedAnalysis::~MotorRainshedAnalysis()
10 10 release();
11 11 }
12 12  
13   -int MotorRainshedAnalysis::init(int devId){
  13 +int MotorRainshedAnalysis::init(int devId, std::string sdk_root){
14 14 ACL_CALL(aclrtCreateContext(&ctx, devId), SY_SUCCESS, SY_FAILED);
15 15  
  16 + std::string model_path = sdk_root + "/models/rainshed/motor_rainshed_231123_310p.om";
  17 +
16 18 mrc_param param;
17   - param.modelNames = "./models/rainshed/motor_rainshed_231123_310p.om";
  19 + param.modelNames = (char*)model_path.data();
18 20 param.thresld = 0.0;
19 21 param.devId = devId;
20 22  
... ...
src/ai_engine_module/MotorRainshedAnalysis.h
  View file @581a68a
... ... @@ -11,7 +11,7 @@ public:
11 11 MotorRainshedAnalysis(/* args */);
12 12 ~MotorRainshedAnalysis();
13 13  
14   - int init(int devId);
  14 + int init(int devId, std::string sdk_root);
15 15  
16 16 vector<MotorRainshedResult> detect(vector<sy_img> vec_img);
17 17  
... ...
src/ai_engine_module/RoadSegAnalysis.cpp
  View file @581a68a
... ... @@ -12,11 +12,13 @@ RoadSegAnalysis::~RoadSegAnalysis()
12 12 release();
13 13 }
14 14  
15   -int RoadSegAnalysis::init(int devId){
  15 +int RoadSegAnalysis::init(int devId, std::string sdk_root){
16 16 ACL_CALL(aclrtCreateContext(&ctx, devId), SY_SUCCESS, SY_FAILED);
17 17  
  18 + std::string model_path = sdk_root + "/models/road_seg/tzroad_seg240108_310p.om";
  19 +
18 20 rs_param param;
19   - param.modelNames = "./models/road_seg/tzroad_seg240108_310p.om";
  21 + param.modelNames = (char*)model_path.data();
20 22 param.thresld = 0.25;
21 23 param.devId = devId;
22 24  
... ... @@ -93,33 +95,20 @@ RoadInfo RoadSegAnalysis::parse_road(rs_result one_result, sy_img src)
93 95 cv::Mat seg_output = seg_post_process(large_resolution, one_result.seg_array, combined, poly_masks, region_classes, lanes, cats, x_sort); //m_masks:mask前的结果 poly_masks后的结果
94 96  
95 97 for (int i = 0; i < lanes.size(); ++i) {
96   - std::vector<cv::Point> points = lanes[i];
97 98 LineInfo info;
98   - for (size_t j = 0; j < points.size(); j++)
99   - {
100   - sy_point pt;
101   - pt.x_ = points[j].x;
102   - pt.y_ = points[j].y;
103   -
104   - info.vec_pt.push_back(pt);
105   - }
  99 + info.vec_pt = lanes[i];
106 100 info.line_type = cats[i];
107 101 one_road.vec_line.push_back(info);
108 102 }
109 103  
110 104 for (int i = 0; i < poly_masks.size(); ++i) {
111   - std::vector<cv::Point> points = poly_masks[i];
112 105 SegInfo seg_info;
113   - for (int j = 0; j < points.size(); ++j) {
114   - sy_point pt;
115   - pt.x_ = points[j].x;
116   - pt.y_ = points[j].y;
117   -
118   - seg_info.vec_pt.push_back(pt);
119   - }
  106 + seg_info.vec_pt = poly_masks[i];
120 107 seg_info.seg_type = region_classes[i];
121 108 one_road.vec_road.push_back(seg_info);
122 109 }
  110 +
  111 + return one_road;
123 112 }
124 113  
125 114 std::vector<SegInfo> RoadSegAnalysis::parse_direct(rs_result one_result, sy_img src) {
... ... @@ -133,21 +122,26 @@ std::vector&lt;SegInfo&gt; RoadSegAnalysis::parse_direct(rs_result one_result, sy_img
133 122  
134 123 cv::Mat seg_output = seg_post_process(large_resolution, one_result.direct_seg, combined, poly_masks, region_classes, lanes, cats, x_sort); //m_masks:mask前的结果 poly_masks后的结果
135 124  
  125 +
  126 + cv::Mat image(cv::Size(640, 360), CV_8UC3, cv::Scalar(0, 0, 0));
  127 +
136 128 std::vector<SegInfo> vec_road;
137 129 for (int i = 0; i < poly_masks.size(); ++i) {
138   - std::vector<cv::Point> points = poly_masks[i];
139 130 SegInfo seg_info;
140   - for (int j = 0; j < points.size(); ++j) {
141   - sy_point pt;
142   - pt.x_ = points[j].x;
143   - pt.y_ = points[j].y;
144   -
145   - seg_info.vec_pt.push_back(pt);
146   - }
147 131 seg_info.seg_type = region_classes[i];
  132 + seg_info.vec_pt = poly_masks[i];
  133 +
  134 + int k = seg_info.seg_type;
  135 + const cv::Scalar color(seg_colors[k][0], seg_colors[k][1], seg_colors[k][2]);
  136 +
  137 + polylines(image, seg_info.vec_pt, true, color, 3, cv::LINE_AA);
  138 +
148 139 vec_road.push_back(seg_info);
149 140 }
150 141  
  142 + cv::imwrite("./direct.jpg", image);
  143 + image.release();
  144 +
151 145 return vec_road;
152 146 }
153 147  
... ... @@ -202,17 +196,15 @@ int RoadSegAnalysis::check_reverse_driving(std::vector&lt;SegInfo&gt;&amp; vec_direct, sy_
202 196 for (size_t i = 0; i < vec_direct.size(); i++) {
203 197 SegInfo& region = vec_direct[i];
204 198 if (region.seg_type == 1) {
205   - std::vector<cv::Point> vec_pt;
206   - for (size_t j = 0; j < region.vec_pt.size(); j++) {
207   - double dist = pointPolygonTest(polygon_pts, cv::Point2f(region.vec_pt[j].x_, region.vec_pt[j].y_), false);
  199 + for (size_t j = 0; j < polygon_pts.size(); j++) {
  200 + double dist = pointPolygonTest(region.vec_pt, polygon_pts[j], false);
208 201 if (dist > 0) {
209 202 coming_count ++;
210 203 }
211 204 }
212 205 } else if (region.seg_type == 2) {
213   - std::vector<cv::Point> vec_pt;
214   - for (size_t j = 0; j < region.vec_pt.size(); j++) {
215   - double dist = pointPolygonTest(polygon_pts, cv::Point2f(region.vec_pt[j].x_, region.vec_pt[j].y_), false);
  206 + for (size_t j = 0; j < polygon_pts.size(); j++) {
  207 + double dist = pointPolygonTest(region.vec_pt, polygon_pts[j], false);
216 208 if (dist > 0) {
217 209 leaving_count ++;
218 210 }
... ... @@ -259,10 +251,10 @@ int RoadSegAnalysis::check_reverse_driving(std::vector&lt;SegInfo&gt;&amp; vec_direct, sy_
259 251 return -1;
260 252 }
261 253  
262   -int RoadSegAnalysis::check_cross_line(std::vector<LineInfo>& vec_line, sy_rect src_rc) {
  254 +int RoadSegAnalysis::check_cross_line(std::vector<LineInfo>& vec_line, sy_rect src_rc, int src_width, int src_height) {
263 255  
264   - float scale_w = 640.0 / src_rc.width_;
265   - float scale_h = 360.0 / src_rc.height_;
  256 + float scale_w = 640.0 / src_width;
  257 + float scale_h = 360.0 / src_height;
266 258  
267 259 sy_rect rc;
268 260 rc.left_ = src_rc.left_ * scale_w;
... ... @@ -294,9 +286,8 @@ int RoadSegAnalysis::check_cross_line(std::vector&lt;LineInfo&gt;&amp; vec_line, sy_rect s
294 286 if (line.line_type == 1 || line.line_type == 2) {
295 287 // 黄实线
296 288 int in_count = 0;
297   - std::vector<cv::Point> vec_pt;
298 289 for (size_t j = 0; j < line.vec_pt.size(); j++) {
299   - double dist = pointPolygonTest(polygon_pts, cv::Point2f(line.vec_pt[j].x_, line.vec_pt[j].y_), false);
  290 + double dist = pointPolygonTest(polygon_pts, line.vec_pt[j], false);
300 291 if (dist > 0) {
301 292 in_count ++;
302 293 }
... ... @@ -312,7 +303,16 @@ int RoadSegAnalysis::check_cross_line(std::vector&lt;LineInfo&gt;&amp; vec_line, sy_rect s
312 303 return -1;
313 304 }
314 305  
315   -int RoadSegAnalysis::check_cross_region(std::vector<SegInfo>& vec_reg, sy_rect rc, int region_type) {
  306 +int RoadSegAnalysis::check_cross_region(std::vector<SegInfo>& vec_reg, sy_rect src_rc, int src_width, int src_height, int region_type) {
  307 +
  308 + float scale_w = 640.0 / src_width;
  309 + float scale_h = 360.0 / src_height;
  310 +
  311 + sy_rect rc;
  312 + rc.left_ = src_rc.left_ * scale_w;
  313 + rc.width_ = src_rc.width_ * scale_w;
  314 + rc.top_ = src_rc.top_ * scale_h;
  315 + rc.height_ = src_rc.height_ * scale_h;
316 316  
317 317 std::vector<cv::Point> polygon_pts;
318 318 cv::Point pt_lt;
... ... @@ -336,11 +336,11 @@ int RoadSegAnalysis::check_cross_region(std::vector&lt;SegInfo&gt;&amp; vec_reg, sy_rect r
336 336 for (size_t i = 0; i < vec_reg.size(); i++) {
337 337 SegInfo& seg = vec_reg[i];
338 338 if (seg.seg_type == region_type) {
339   - // 黄实线
  339 +
340 340 int in_count = 0;
341   - std::vector<cv::Point> vec_pt;
  341 + // 车辆与region相交
342 342 for (size_t j = 0; j < seg.vec_pt.size(); j++) {
343   - double dist = pointPolygonTest(polygon_pts, cv::Point2f(seg.vec_pt[j].x_, seg.vec_pt[j].y_), false);
  343 + double dist = pointPolygonTest(polygon_pts, seg.vec_pt[j], false);
344 344 if (dist > 0) {
345 345 in_count ++;
346 346 }
... ... @@ -350,76 +350,20 @@ int RoadSegAnalysis::check_cross_region(std::vector&lt;SegInfo&gt;&amp; vec_reg, sy_rect r
350 350 {//有5个点就认为是压线了
351 351 return 1;
352 352 }
  353 +
  354 + // 车辆与region相交情形未检测出来，检测车辆在region中情形
  355 + for (size_t j = 0; j < polygon_pts.size(); j++) {
  356 + double dist = pointPolygonTest(seg.vec_pt, polygon_pts[j], false);
  357 + if (dist > 0) {
  358 + return 1;
  359 + }
  360 + }
353 361 }
354 362 }
355 363  
356 364 return -1;
357 365 }
358 366  
359   -// cv::Mat imshow_lanes(cv::Mat img, const rs_lane* lanes, int lane_count) {
360   -// float scale_w = img.cols / 640.0;
361   -// float scale_h = img.rows / 360.0;
362   -// std::vector<std::pair<std::vector<cv::Point>, int>> combined;
363   -// lanes_process(lanes, lane_count, combined, scale_w, scale_h);
364   -
365   -// for (const auto& lane_info : combined) {
366   -// const auto& xys = lane_info.first;
367   -// int cls = lane_info.second;
368   -// cv::Scalar color(lane_colors[cls][0],lane_colors[cls][1],lane_colors[cls][2]);
369   -// for (size_t i = 1; i < xys.size(); ++i) {
370   -// cv::line(img, xys[i - 1], xys[i], color, 4);
371   -// }
372   -// }
373   -
374   -// return img;
375   -// }
376   -
377   -// void RoadSegAnalysis::post_direct(rs_result one_result, sy_img src){
378   -// int src_width = src.w_;
379   -// int src_height = src.h_;
380   -// int w = 640;
381   -// int h = 360;
382   -// float alpha = 0.75;
383   -// cv::Mat overlayed_direct_img(cv::Size(w,h), CV_8UC3, cv::Scalar(0, 0, 0));
384   -// // 将车道线标签转换为彩色图像
385   -// for (int i = 0; i < h; ++i) {
386   -// for (int j = 0; j < w; ++j) {
387   -// int idx = one_result.direct_seg[(i * w + j)];
388   -// overlayed_direct_img.at<cv::Vec3b>(i, j)[0] = seg_colors[idx][0]; // R通道;
389   -// overlayed_direct_img.at<cv::Vec3b>(i, j)[1] = seg_colors[idx][1]; // G通道
390   -// overlayed_direct_img.at<cv::Vec3b>(i, j)[2] = seg_colors[idx][2]; // B通道
391   -// }
392   -// }
393   -// cv::resize(overlayed_direct_img, overlayed_direct_img, cv::Size(src_width,src_height), 0, 0, cv::INTER_LINEAR);
394   -// // 将原始图像和彩色车道线图进行混合
395   -// // cv::addWeighted(cvImg, alpha, overlayed_direct_img, 1 - alpha, 0, overlayed_direct_img);
396   -// cv::Mat img_direct_lane = imshow_lanes(overlayed_direct_img, one_result.reg_array, one_result.lane_count);
397   -// cv::imwrite("img_direct_lane.jpg", img_direct_lane);
398   -
399   -
400   -// std::vector<std::pair<std::vector<cv::Point>, int>> combined;
401   -// lanes_process(one_result.reg_array, one_result.lane_count, combined);
402   -// std::vector<std::vector<cv::Point>> poly_masks, lanes, direct_masks, merge_masks;
403   -// std::vector<int> region_classes, cats, direct_classes, merge_classes;
404   -// cv::Mat background_mask;
405   -// std::map<double, int> x_sort;
406   -// bool large_resolution = false;
407   -// if (src_height > 1920) large_resolution = true;
408   -// cv::Mat seg_output = seg_post_process(large_resolution, one_result.seg_array, combined, poly_masks, region_classes, lanes, cats, x_sort, background_mask); //m_masks:mask前的结果 poly_masks后的结果
409   -// cv::Mat direct_output = direct_post_process(large_resolution, one_result.direct_seg, direct_masks, direct_classes);
410   -// cv::Mat merge_output = merge_direct_process(large_resolution, one_result.direct_seg, merge_masks, merge_classes);
411   -
412   -// cv::Mat image(cv::Size(w,h), CV_8UC3, cv::Scalar(0, 0, 0));
413   -// cv::Mat vis_image = mask_to_rgb(image, seg_output);
414   -// cv::Mat direct_image = mask_to_rgb(image, direct_output);
415   -// cv::Mat merge_image = mask_to_rgb(image, merge_output);
416   -// cv::imwrite("vis_image.jpg", vis_image);
417   -// cv::imwrite("direct_image.jpg", direct_image);
418   -// cv::imwrite("merge_image.jpg", merge_image);
419   -// cv::imwrite("background_mask.jpg", background_mask);
420   -// }
421   -
422   -
423 367 cv::Mat RoadSegAnalysis::mask_to_rgb(cv::Mat img, cv::Mat mask) {
424 368 cv::Mat masks = img.clone();
425 369 int reg_cls = 8;
... ... @@ -612,10 +556,6 @@ cv::Mat RoadSegAnalysis::seg_post_process(bool large_resolution, unsigned char *
612 556 if (x_sort.count(centr_x)) centr_x += 0.0001;
613 557 x_sort.insert(std::make_pair(centr_x, count));
614 558 ++ count;
615   - // for (auto iter: contours_poly) {
616   - // std::cout << "contour " << iter << " " << cat << std::endl;
617   - // }
618   -
619 559 }
620 560 }
621 561 }
... ...
src/ai_engine_module/RoadSegAnalysis.h
  View file @581a68a
... ... @@ -18,15 +18,15 @@ public:
18 18 RoadSegAnalysis(/* args */);
19 19 ~RoadSegAnalysis();
20 20  
21   - int init(int devId);
  21 + int init(int devId, std::string sdk_root);
22 22  
23 23 std::vector<RoadInfo> detect(vector<sy_img> vec_img);
24 24  
25 25 int check_reverse_driving(std::vector<SegInfo>& vec_direct, sy_rect rc, int src_width, int src_height, int head_or_tail);
26 26  
27   - int check_cross_line(std::vector<LineInfo>& vec_line, sy_rect rc);
  27 + int check_cross_line(std::vector<LineInfo>& vec_line, sy_rect rc, int src_width, int src_height);
28 28  
29   - int check_cross_region(std::vector<SegInfo>& vec_reg, sy_rect rc, int region_type);
  29 + int check_cross_region(std::vector<SegInfo>& vec_reg, sy_rect rc, int src_width, int src_height, int region_type);
30 30  
31 31 private:
32 32 int release();
... ...
src/ai_engine_module/VehicleAnalysis.cpp
  View file @581a68a
... ... @@ -9,7 +9,7 @@ VehicleAnalysis::~VehicleAnalysis() {
9 9 release();
10 10 }
11 11  
12   -int VehicleAnalysis::init(int devId, int max_batch_size) {
  12 +int VehicleAnalysis::init(int devId, std::string sdk_root, int max_batch_size) {
13 13  
14 14 param.vehicle_detect_config= SY_CONFIG_OPEN; //1.开启车检测 SY_CONFIG_CLOSE SY_CONFIG_OPEN
15 15 param.vehicle_recg_config= SY_CONFIG_OPEN; //4.开启车型识别
... ... @@ -52,9 +52,11 @@ int VehicleAnalysis::init(int devId, int max_batch_size) {
52 52 //车颜色阈值
53 53 param.vc_thresld = 0.5;
54 54  
  55 + string dbPath = sdk_root + "/models/vehicle_analysis/db/vr_h0725x210605_r191230.db";
  56 + string models_path = sdk_root + "/models/vehicle_analysis/";
55 57 //车型参数
56   - param.dbPath="./models/vehicle_analysis/db/vr_h0725x210605_r191230.db";
57   - param.models_Path="./models/vehicle_analysis/"; //所有模型的地址
  58 + param.dbPath= (char*)dbPath.data();
  59 + param.models_Path= (char*)models_path.data(); //所有模型的地址
58 60  
59 61 cout << "va_init start " << endl;
60 62 // 内部有 ctx
... ...
src/ai_engine_module/VehicleAnalysis.h
  View file @581a68a
... ... @@ -38,7 +38,7 @@ public:
38 38 VehicleAnalysis(/* args */);
39 39 ~VehicleAnalysis();
40 40  
41   - int init(int devId, int max_batch_size);
  41 + int init(int devId, std::string sdk_root, int max_batch_size);
42 42  
43 43 va_result* detect(vector<sy_img> vec_img);
44 44  
... ...
src/ai_engine_module/VehicleHeadTail.cpp
  View file @581a68a
... ... @@ -17,8 +17,10 @@ int VehicleHeadTail::init(head_tail_param param){
17 17  
18 18 LOG_INFO("head_tail version: head_tail_256.2024.12.18");
19 19  
  20 + string model_path = param.sdk_root + "/models/car_head_tail/head_tail_256_241220_310P.om";
  21 +
20 22 m_cnn_cls = new CnnCls();
21   - int ret = m_cnn_cls->Init("./models/car_head_tail/head_tail_256_241220_310P.om");
  23 + int ret = m_cnn_cls->Init(model_path.c_str());
22 24 if (ret != SY_SUCCESS) {
23 25 delete m_cnn_cls;
24 26 m_cnn_cls = nullptr;
... ...
src/ai_engine_module/VehicleHeadTail.h
  View file @581a68a
... ... @@ -14,6 +14,7 @@ typedef struct head_tail_param
14 14 {
15 15 int devId; //ָ指定显卡id
16 16 int max_batch;
  17 + std::string sdk_root;
17 18 head_tail_param() :devId(0), max_batch(8){};
18 19 } head_tail_param;
19 20  
... ...
src/ai_engine_module/VidClothes.cpp
  View file @581a68a
... ... @@ -9,11 +9,13 @@ VidClothes::~VidClothes()
9 9 release();
10 10 }
11 11  
12   -int VidClothes::init(int devId){
  12 +int VidClothes::init(int devId, std::string sdk_root){
13 13 ACL_CALL(aclrtCreateContext(&ctx, devId), SY_SUCCESS, SY_FAILED);
14 14  
  15 + std::string model_path = sdk_root + "/models/vid_clothes/vidClothes0325_310P.om";
  16 +
15 17 vidclothes_param param;
16   - param.modelNames = "./models/vid_clothes/vidClothes0325_310P.om";
  18 + param.modelNames = (char*)model_path.data();
17 19 param.thresld = 0.0;
18 20 param.devId = devId;
19 21  
... ...
src/ai_engine_module/VidClothes.h
  View file @581a68a
... ... @@ -7,7 +7,7 @@ public:
7 7 VidClothes(/* args */);
8 8 ~VidClothes();
9 9  
10   - int init(int devId);
  10 + int init(int devId, std::string sdk_root);
11 11  
12 12 vector<int> detect(vector<sy_img> vec_img);
13 13  
... ...
src/ai_engine_module/va_test.cpp0 deleted
View file @8d31977
1   -#include <iostream>
2   -#include <sstream>
3   -#include <fstream>
4   -
5   -#include "sy_common.h"
6   -#include "sy_errorinfo.h"
7   -#include "vehicle_analysis.h"
8   -#include "vehicle_result.h"
9   -
10   -#include "opencv2/opencv.hpp"
11   -#include "opencv2/imgcodecs/legacy/constants_c.h"
12   -#include "opencv2/imgproc/types_c.h"
13   -#include "time.h"
14   -#include "sys/time.h"
15   -
16   -#include "utils.h"
17   -#include "dvpp_process.h"
18   -
19   -using namespace std;
20   -using namespace cv;
21   -
22   -#include <chrono>
23   -#include <dirent.h>
24   -
25   -
26   -
27   -//static char colorLabel[13][8] = { "棕", "橙", "灰", "白", "粉", "紫", "红", "绿", "蓝", "银", "青", "黄", "黑" };
28   -static std::vector<std::string> CARCOLOR1 = {
29   - "BROWN","ORANGE","GRAY","WHITE","PINK","PURPLE","RED","GREEN","BLUE",
30   - "SILVER","CYAN","YELLOW","BLACK"
31   -};
32   -
33   -static char vsLabel[11][40] = { "diao", "guan", "huo", "zha", "jiao","jiu", "la", "xiao", "qing", "sui","wei" };
34   -
35   -
36   -double msecond() {
37   - struct timeval tv;
38   - gettimeofday(&tv, 0);
39   - return (tv.tv_sec * 1000.0 + tv.tv_usec / 1000.0);
40   -}
41   -
42   -
43   -void getAllNm(std::string pthNm, std::vector<std::string>& fileList)
44   -{
45   - DIR *dir;
46   - struct dirent *ptr;
47   - dir = opendir(pthNm.c_str()); ///open the dir
48   - int filenum = 0;
49   - while((ptr = readdir(dir)) != NULL) ///read the list of this dir
50   - {
51   - // char* to string
52   - std::string curNm = ptr->d_name;
53   - if(curNm != "." && curNm != "..")
54   - {
55   - filenum++;
56   - fileList.push_back(curNm);
57   - //printf("file %d name: %s\n", filenum, curNm.c_str());
58   - }
59   - }
60   - closedir(dir);
61   -}
62   -
63   -
64   -void CvBGR2NV21(Mat& bgr, unsigned char* yuv) {
65   - int stride = (bgr.cols + 127) / 128 * 128;
66   - int strideH = (bgr.rows + 15) / 16 * 16;
67   - for (int i = 0; i < bgr.rows; i++) {
68   - for (int j = 0; j < bgr.cols; j++) {
69   - int B = bgr.at<cv::Vec3b>(i, j)[0];
70   - int G = bgr.at<cv::Vec3b>(i, j)[1];
71   - int R = bgr.at<cv::Vec3b>(i, j)[2];
72   -
73   - int Y = (77 * R + 150 * G + 29 * B) >> 8;
74   - yuv[i * stride + j] = (Y < 0) ? 0 : ((Y > 255) ? 255 : Y);
75   - if (i % 2 == 0 && j % 2 == 0) {
76   - int U = ((-44 * R - 87 * G + 131 * B) >> 8) + 128;
77   - int V = ((131 * R - 110 * G - 21 * B) >> 8) + 128;
78   - yuv[strideH * stride + i / 2 * stride + j] = (V < 0) ? 0 : ((V > 255) ? 255 : V);
79   - yuv[strideH * stride + i / 2 * stride + j + 1] = (U < 0) ? 0 : ((U > 255) ? 255 : U);
80   - }
81   - }
82   - }
83   -}
84   -
85   -
86   -
87   -
88   -
89   -
90   -
91   -int main() {
92   -
93   - //ofstream oFile;
94   - //打开要输出的文件
95   - //oFile.open("va_result.csv", ios::out | ios::trunc);
96   - //oFile << "image" << "," << "image_quality_score" << "," << "count" << "," << "vehicle_info" << endl;
97   -
98   -
99   - cout << va_get_version() << endl;
100   -
101   - //string imagepath= "./vpt/";
102   - //string saveimagepath= "./result/";
103   - //int gpuid = 0;
104   -
105   - //string imagepath= "/home/HwHiAiUser/WH/vehicle_analysis202105/data/vpt/";
106   - //string imagepath= "/home/HwHiAiUser/WH/vehicle_analysis202105/data/bug/";
107   - string img_file_path= "/data/wanghong/Atlas310pro/vehicle_analysis202303/data/bug/";
108   - //string imagepath= "/home/HwHiAiUser/WH/vehicle_analysis202105/data/img/";
109   - //string imagepath= "/home/HwHiAiUser/WH/vehicle_analysis202105/data/carset/";
110   - string saveimagepath= "/data/wanghong/Atlas310pro/vehicle_analysis202303/data/result/";
111   - int gpuid = 0;
112   -
113   - va_param param;
114   - param.vehicle_detect_config= SY_CONFIG_OPEN; //1.开启车检测 SY_CONFIG_CLOSE SY_CONFIG_OPEN
115   - param.vehicle_color_config= SY_CONFIG_OPEN; //3.开启车颜色识别
116   - param.vehicle_recg_config= SY_CONFIG_OPEN; //4.开启车型识别
117   - param.vehicle_recg_supplement_config= SY_CONFIG_OPEN; //4.开启车型识别补充识别
118   - param.vehicle_plate_det_recg_config= SY_CONFIG_OPEN; //5.开启车牌检测识别
119   - param.vehicle_pendant_det_config= SY_CONFIG_OPEN; //6.开启车属性检测识别
120   - param.vehicle_illegal_config= SY_CONFIG_OPEN; //7.开启车违规
121   - param.vehicle_feature_config= SY_CONFIG_OPEN; //8.开启车辆特征提取
122   - param.vehicle_motor_tricycle_analysis_config= SY_CONFIG_OPEN; //8.摩托车分析
123   - param.vehicle_manned_config= SY_CONFIG_OPEN; //8.开启载人分析
124   -
125   -
126   - param.gpuid=gpuid;
127   -
128   - //车检测参数
129   - //param.vehicle_det_param.process_min_l = 720;// 720;
130   - //param.vehicle_det_param.process_max_l = 1280;//1280;
131   - //param.vehicle_det_param.thresld=0.3;
132   - //param.min_obj_size=200; //最小目标
133   - param.vehicle_det_thresld=0.4;
134   -
135   - //车牌检测参数
136   - //param.vehicle_plate_det_param.process_min_l=320;
137   - //param.vehicle_plate_det_param.process_max_l=320;
138   - //param.vehicle_plate_det_param.thresld=0.4;
139   - param.vehicle_plate_det_thresld=0.5;
140   -
141   - //车属性检测参数
142   - //param.vehicle_attribute_det_param.process_min_l=360;
143   - //param.vehicle_attribute_det_param.process_max_l=640;
144   - //param.vehicle_attribute_det_param.thresld=0.3;
145   - param.vehicle_attribute_det_thresld=0.5;
146   -
147   - //车logo检测参数
148   - //param.vehicle_logo_det_param.process_min_l=512;
149   - //param.vehicle_logo_det_param.process_max_l=512;
150   - //param.vehicle_logo_det_param.thresld=0.1;
151   - param.vehicle_logo_det_thresld=0.1;
152   -
153   - //车颜色阈值
154   - param.vc_thresld = 0.5;
155   -
156   - //车型参数
157   - param.dbPath="/data/wanghong/Atlas310pro/vehicle_analysis202303/vehicle_analysis/db/vr_h0725x210605_r191230.db";
158   - //param.dbPath="../db/vr_h0725x210605_r191230.db";
159   -
160   - param.models_Path="/data/wanghong/Atlas310pro/vehicle_analysis202303/vehicle_analysis/all_models/"; //所有模型的地址
161   - //param.models_Path="../modelsPath/"; //所有模型的地址
162   -
163   -
164   - int ret0= va_acl_init();
165   -
166   - void* handle = nullptr;
167   - cout << "init start " << endl;
168   - int ret = va_init(&handle, param);
169   - if (ret == 0) {
170   - cout << "init success " << endl;
171   - }
172   - else
173   - return ret;
174   -
175   -
176   -
177   -
178   -// void* handle111 = nullptr;
179   -// cout << "init2 start " << endl;
180   -// int ret2 = va_init(&handle111, param);
181   -// if (ret2 == 0) {
182   -// cout << "init2 success " << endl;
183   -// }
184   -// else
185   -// {
186   -// cout << "init2 failed " << endl;
187   -// return ret2;
188   -// }
189   -
190   -
191   -
192   - double t3=0;
193   - int count =0;
194   -
195   -//temp
196   - aclrtContext ctx;
197   - ACL_CALL(aclrtCreateContext(&ctx, gpuid), ACL_ERROR_NONE, SY_FAILED);
198   - ACL_CALL(aclrtSetCurrentContext(ctx), ACL_ERROR_NONE, SY_FAILED);
199   -
200   - aclrtStream stream = nullptr;
201   - ACL_CALL(aclrtCreateStream(&stream), ACL_ERROR_NONE, SY_FAILED);
202   - DvppProcess* dvpp = new DvppProcess();
203   - dvpp->InitResource(stream);
204   -
205   -for(int iter =0;iter<100;iter++)
206   -{
207   - //ifstream infile("list.txt");
208   - //ifstream infile("list_bug.txt");
209   - //ifstream infile("list_700W.txt");
210   - //ifstream infile("list_img.txt");
211   - //string file;
212   - std::vector<std::string> fileList;
213   - getAllNm(img_file_path,fileList);
214   - if (fileList.empty()) throw std::logic_error("No suitable images were found");
215   -
216   -
217   - for (auto & file : fileList)
218   - {
219   - string filename = img_file_path + file;
220   - cout << "img path: " << filename << endl;
221   - const int batch_size = 1;
222   - printf("batch_size=%d \n",batch_size);
223   -
224   - Mat cvImg = imread(filename.c_str());
225   - sy_img imgs[batch_size];
226   -
227   - ACL_CALL(aclrtSetCurrentContext(ctx), ACL_ERROR_NONE, SY_FAILED);
228   - ImageData src[batch_size], dvpp_data[batch_size];
229   - if(1)
230   - {
231   - for (int b = 0; b < batch_size; b++) {
232   - Utils::ReadImageFile(src[b], filename); //将二进制图像读入内存，并读取宽高信息
233   - ACL_CALL(dvpp->CvtJpegToYuv420sp(dvpp_data[b], src[b]), SY_SUCCESS, SY_FAILED); //解码
234   - imgs[b].w_ = dvpp_data[b].width;//dvpp_data[b].alignWidth;
235   - imgs[b].h_ = dvpp_data[b].height;//dvpp_data[b].alignHeight;
236   - imgs[b].data_ = dvpp_data[b].data.get();//(uint8_t*)Utils::CopyDataDeviceToLocal(dvpp_data[b].data.get(), dvpp_data[b].size);
237   - }
238   - }
239   -
240   - ImageData img_data_dvpp[batch_size];
241   - sy_img imgs_device[batch_size];
242   - if(0)
243   - {
244   - uint32_t alignWidth = (cvImg.cols + 127) / 128 * 128;
245   - uint32_t alignHeight = (cvImg.rows + 15) / 16 * 16;
246   - uint32_t size = alignWidth * alignHeight * 1.5;
247   - std::shared_ptr<uint8_t> data = shared_ptr<uint8_t>((new uint8_t[size]),
248   - [](uint8_t* p) {delete [] p;});
249   - CvBGR2NV21(cvImg, data.get());
250   -
251   - for (int i = 0; i < batch_size; i++) {
252   - imgs[i].w_ = cvImg.cols;
253   - imgs[i].h_ = cvImg.rows;
254   - imgs[i].data_ = data.get();
255   - //转到device输入
256   - Utils::CopysyImageDataToDvpp(img_data_dvpp[i], imgs[i]);
257   - imgs_device[i].w_ = img_data_dvpp[i].width;//dvpp_data[b].alignWidth;
258   - imgs_device[i].h_ = img_data_dvpp[i].height;//dvpp_data[b].alignHeight;
259   - imgs_device[i].data_ = img_data_dvpp[i].data.get();
260   - }
261   - }
262   -
263   - if(0)
264   - {
265   - uint32_t alignWidth = (cvImg.cols + 127) / 128 * 128;
266   - uint32_t alignHeight = (cvImg.rows + 15) / 16 * 16;
267   - uint32_t size = alignWidth * alignHeight * 1.5;
268   - std::shared_ptr<uint8_t> data = shared_ptr<uint8_t>((new uint8_t[size]),
269   - [](uint8_t* p) {delete [] p;});
270   - CvBGR2NV21(cvImg, data.get());
271   -
272   - for (int i = 0; i < batch_size; i++) {
273   - imgs[i].w_ = cvImg.cols;
274   - imgs[i].h_ = cvImg.rows;
275   - imgs[i].data_ = data.get();
276   - }
277   -
278   - }
279   -
280   - count++;
281   -
282   - va_result *result=new va_result[batch_size];
283   - for(int b=0;b<batch_size;b++)
284   - {
285   - result[b].count=0;
286   - result[b].info=new vehicle_info[100];
287   - for(int c=0; c<100; c++)
288   - {
289   - result[b].info[c].vehicle_pendant_det_res.vpd_res=new v_pendant_d_info[300];
290   - }
291   - }
292   -
293   - double t1,t2;
294   - t1 = msecond();
295   -
296   - std::cout << "Predict----------------------------------"<< std::endl;
297   - int ret = va_batch(handle, imgs, batch_size, result);
298   - //int ret = va_batch(handle, imgs_device, batch_size, result);
299   - t2 = msecond();
300   - printf("va_teat info:va_batch time: %.2f\n", (t2 - t1));
301   - t3 += t2-t1;
302   -
303   - std::cout << "output----------------------------------"<< std::endl;
304   -
305   - for (int b = 0; b < batch_size; b++)
306   - {
307   - if(param.vehicle_detect_config==SY_CONFIG_CLOSE)//不做车头检测时，应设result[b].count=1，即输入是车头图像，只含1辆车。
308   - result[b].count=1;
309   -
310   - //oFile << file << ","<< result[b].count<< "," <<endl;;
311   - std::cout <<"car_count:" <<result[b].count<< std::endl;
312   - for(int c=0;c<result[b].count;c++)
313   - {
314   - //temp，中间结果，输出车牌被置黑的车小图
315   - //if(0)
316   - //{
317   - // cv::Mat tmp(result[b].info[c].car_img_data_cpu[0].h_, result[b].info[c].car_img_data_cpu[0].w_, CV_8UC3, result[b].info[c].car_img_data_cpu[0].data_);
318   - // char str_i[100];
319   - // sprintf(str_i, "%d_", c);
320   - // cv::imwrite((save_folder_roi+str_i+i).c_str(),tmp);
321   - //}
322   -
323   -
324   - std::string str_vehicle_type;
325   - int vehicle_type=result[b].info[c].type;
326   - //std::cout << "vehicle_type="<<vehicle_type<< std::endl;
327   - if (vehicle_type==0)str_vehicle_type = "head";
328   - else if (vehicle_type==1)str_vehicle_type = "rear";
329   - else if (vehicle_type==2)str_vehicle_type = "motor";
330   - else if (vehicle_type==3)str_vehicle_type = "tricycle";
331   - else if (vehicle_type==4)str_vehicle_type = "body";
332   - else if (vehicle_type==5)str_vehicle_type = "body_nova";
333   - std::cout << " vehicle_type:"<<str_vehicle_type<< std::endl;
334   -
335   - //1.车头结果：蓝框//2.车窗结果：绿框
336   - if(param.vehicle_detect_config==SY_CONFIG_OPEN)
337   - {
338   - //车身
339   - //std::cout << "vehicle_body_detect_res info:"<< std::endl;
340   - if(result[b].info[c].vehicle_body_detect_res.rect.width_>0)
341   - {
342   - float vehicle_body_detect_res_score = result[b].info[c].vehicle_body_detect_res.score;
343   - int x1=result[b].info[c].vehicle_body_detect_res.rect.left_;
344   - int y1=result[b].info[c].vehicle_body_detect_res.rect.top_;
345   - int x2=result[b].info[c].vehicle_body_detect_res.rect.left_+result[b].info[c].vehicle_body_detect_res.rect.width_;
346   - int y2=result[b].info[c].vehicle_body_detect_res.rect.top_+result[b].info[c].vehicle_body_detect_res.rect.height_;
347   - std::cout << " vehicle_body_detect_res_score:" <<vehicle_body_detect_res_score<<",car_rect:[" <<x1<<"," <<y1<<"," <<x2-x1<<"," <<y2-y1<<"]"<< std::endl;
348   -
349   - Point lt(x1, y1);
350   - Point rb(x2, y2);
351   - rectangle(cvImg, lt, rb, cv::Scalar(0, 255,255), 4);
352   - }
353   -
354   - //车头
355   - //std::cout << "vehicle_detect_res info:"<< std::endl;
356   - if(result[b].info[c].vehicle_detect_res.rect.width_>0)
357   - {
358   - float vehicle_detect_res_score = result[b].info[c].vehicle_detect_res.score;
359   - int x1=result[b].info[c].vehicle_detect_res.rect.left_;
360   - int y1=result[b].info[c].vehicle_detect_res.rect.top_;
361   - int x2=result[b].info[c].vehicle_detect_res.rect.left_+result[b].info[c].vehicle_detect_res.rect.width_;
362   - int y2=result[b].info[c].vehicle_detect_res.rect.top_+result[b].info[c].vehicle_detect_res.rect.height_;
363   - std::cout << " vehicle_detect_res_score:" <<vehicle_detect_res_score<<",car_rect:[" <<x1<<"," <<y1<<"," <<x2-x1<<"," <<y2-y1<<"]"<< std::endl;
364   -
365   - Point lt(x1, y1);
366   - Point rb(x2, y2);
367   - rectangle(cvImg, lt, rb, cv::Scalar(255, 0, 0), 4);
368   - }
369   -
370   - //车窗
371   - //std::cout << "vwd output----"<< std::endl;
372   - if(result[b].info[c].vehicle_win_detect_res.rect.width_>0)
373   - {
374   - float vehicle_win_detect_res_score = result[b].info[c].vehicle_win_detect_res.score;
375   - int x1=result[b].info[c].vehicle_win_detect_res.rect.left_;
376   - int y1=result[b].info[c].vehicle_win_detect_res.rect.top_;
377   - int x2=result[b].info[c].vehicle_win_detect_res.rect.left_+result[b].info[c].vehicle_win_detect_res.rect.width_;
378   - int y2=result[b].info[c].vehicle_win_detect_res.rect.top_+result[b].info[c].vehicle_win_detect_res.rect.height_;
379   - std::cout << " vehicle_win_detect_res_score:" <<vehicle_win_detect_res_score<<",win_rect:[" <<x1<<"," <<y1<<"," <<x2-x1<<"," <<y2-y1<<"]"<< std::endl;
380   -
381   - Point lt(x1, y1);
382   - Point rb(x2, y2);
383   - rectangle(cvImg, lt, rb, cv::Scalar(0, 255, 0), 4);
384   - }
385   - }
386   -
387   -
388   - //3.车颜色
389   - if(param.vehicle_color_config==SY_CONFIG_OPEN&& (vehicle_type==0 || vehicle_type==1 || vehicle_type==4))
390   - {
391   - //std::cout << "vc output----"<< std::endl;
392   - int index=result[b].info[c].vehicle_color_res.index;
393   - float score=result[b].info[c].vehicle_color_res.score;
394   -
395   - if(score>0.5)
396   - {
397   - std::cout << " car color info:"<< endl;
398   - std::cout << " index:"<< index<<" score:"<< score<< std::endl;
399   -
400   - std::string str_i = CARCOLOR1[index];
401   - cv::putText(cvImg, str_i,
402   - cv::Point(result[b].info[c].vehicle_body_detect_res.rect.left_+5, result[b].info[c].vehicle_body_detect_res.rect.top_+50),
403   - cv::FONT_HERSHEY_COMPLEX_SMALL,
404   - 1.3,
405   - cv::Scalar(0, 0, 255),
406   - 2);
407   - }
408   - }
409   -
410   -
411   - //4.VR车型识别
412   - if(param.vehicle_recg_config==SY_CONFIG_OPEN && (vehicle_type==0 || vehicle_type==1 || vehicle_type==4 ))
413   - {
414   - //std::cout << "vr output----"<< std::endl;
415   -
416   - char *vehicle_brand=result[b].info[c].vehicle_recg_res.vehicle_brand;//车辆品牌
417   - char* vehicle_subbrand=result[b].info[c].vehicle_recg_res.vehicle_subbrand; //车辆子品牌
418   - char* vehicle_issue_year=result[b].info[c].vehicle_recg_res.vehicle_issue_year; //车辆年款
419   - char* vehicle_type_=result[b].info[c].vehicle_recg_res.vehicle_type; //车辆类型
420   - char* freight_ton=result[b].info[c].vehicle_recg_res.freight_ton; //货车吨级
421   - float name_score=result[b].info[c].vehicle_recg_res.name_score; //识别置信度
422   -
423   - float name_score_thre = 0;
424   - //if(vehicle_type==0)name_score_thre=0.7;//车头车型识别建议阈值0.7
425   - //if(vehicle_type==1)name_score_thre=0.8;//车尾车型识别建议阈值0.8
426   - if(name_score > name_score_thre)
427   - {
428   - cout << " vehicle_recg info: " << endl;
429   - std::cout << " name_score: " << name_score << std::endl;
430   - std::cout << " vehicle_brand: " << vehicle_brand << std::endl;
431   - std::cout << " vehicle_subbrand: " << vehicle_subbrand << std::endl;
432   - std::cout << " vehicle_issue_year: " << vehicle_issue_year << std::endl;
433   - std::cout << " vehicle_type_: " << vehicle_type_ << std::endl;
434   - std::cout << " freight_ton: " << freight_ton << std::endl;
435   -
436   - //printf("name_score:%f\n", name_score);
437   - //printf("vehicle_brand:%s\n", vehicle_brand);
438   - //printf("vehicle_subbrand:%s\n", vehicle_subbrand);
439   - //printf("vehicle_issue_year:%s\n", vehicle_issue_year);
440   - //printf("vehicle_type_:%s\n", vehicle_type_);
441   - //printf("freight_ton:%s\n", freight_ton);
442   - }
443   -
444   - }
445   -
446   -
447   - //5.VP车牌检测识别
448   - if(param.vehicle_plate_det_recg_config==SY_CONFIG_OPEN && (vehicle_type==0 || vehicle_type==1))
449   - {
450   - std::cout << " car plate info:"<< endl;
451   - int special_type=result[b].info[c].vehicle_plate_det_recg_res.special_type;//常规车牌、临时车牌、低速车牌。0-common,1-temporary_license_plate,2-low_speed_license_plate.
452   - std::cout << " special_type:" << special_type<< std::endl;
453   -
454   - float detect_score=result[b].info[c].vehicle_plate_det_recg_res.detect_score;
455   - //std::cout << "detect_score:" << detect_score<< std::endl;
456   - if(detect_score>0.3)
457   - {
458   - int type=result[b].info[c].vehicle_plate_det_recg_res.type;
459   - sy_rect rect=result[b].info[c].vehicle_plate_det_recg_res.rect;
460   -
461   - int x1=rect.left_;
462   - int y1=rect.top_;
463   - int x2=rect.left_+rect.width_;
464   - int y2=rect.top_+rect.height_;
465   - std::cout << " vp_type:"<<type<<",vp_det_score:" <<detect_score <<",vp_rect:[" <<x1<<"," <<y1<<"," <<x2-x1<<"," <<y2-y1<<"]"<< std::endl;
466   -
467   - Point lt(x1, y1);
468   - Point rb(x2, y2);
469   - rectangle(cvImg, lt, rb, cv::Scalar(0, 0, 255), 4);
470   -
471   -
472   -
473   - //车牌识别结果
474   - float num_score=result[b].info[c].vehicle_plate_det_recg_res.num_score;
475   - std::string plate_recg="";
476   - //if(num_score>0.99)//车牌识别建议置信度阈值0.99
477   - {
478   -
479   - //printf("plate: ");
480   - for (int m = 0; m < PLATENUM; m++)
481   - {
482   - //printf("%s", result[b].info[c].vehicle_plate_det_recg_res.recg[m].character);
483   - plate_recg=plate_recg + result[b].info[c].vehicle_plate_det_recg_res.recg[m].character;
484   - }
485   - //printf("\n");
486   -
487   - //printf("maxprob: ");
488   - //for (int m = 0; m < PLATENUM; m++)
489   - //{
490   - // printf("%f ", result[b].info[c].vehicle_plate_det_recg_res.vehicle_plate_infos[0].recg[m].maxprob);
491   - //}
492   - //printf("\n");
493   -
494   -
495   - //std::cout << "type:" << type<<" num_score:" <<num_score<<" " << std::endl;
496   - std::cout <<" "<<plate_recg<< ",vp_num_score:" << num_score<< std::endl;
497   -
498   - }
499   -
500   -
501   - char str_cc[100];
502   - memset(str_cc, 0, sizeof(str_cc));
503   - cv::Point2f p1;
504   - p1.x = x1;
505   - p1.y = y1-60;
506   - //sprintf(str_cc, "det_%.2f_recg_%.2f_%s",detectScore,numScore,plate_recg.c_str());
507   - sprintf(str_cc, "det_%.2f",detect_score);
508   - cv::putText(cvImg, str_cc,
509   - p1,
510   - cv::FONT_HERSHEY_COMPLEX_SMALL,
511   - 1.3,
512   - cv::Scalar(0, 0, 255),
513   - 2);
514   -
515   - memset(str_cc, 0, sizeof(str_cc));
516   - p1.x = x1;
517   - p1.y = y1-30;
518   - sprintf(str_cc, "recg_%.2f",num_score);
519   - cv::putText(cvImg, str_cc,
520   - p1,
521   - cv::FONT_HERSHEY_COMPLEX_SMALL,
522   - 1.3,
523   - cv::Scalar(0, 0, 255),
524   - 2);
525   -
526   - memset(str_cc, 0, sizeof(str_cc));
527   - p1.x = x1;
528   - p1.y = y1;
529   - sprintf(str_cc, "%s",plate_recg.c_str());
530   - cv::putText(cvImg, str_cc,
531   - p1,
532   - cv::FONT_HERSHEY_COMPLEX_SMALL,
533   - 1.3,
534   - cv::Scalar(0, 0, 255),
535   - 2);
536   -
537   - }
538   -
539   - }
540   -
541   - //6.车属性结果：
542   - if(param.vehicle_pendant_det_config==SY_CONFIG_OPEN && (vehicle_type==0|| vehicle_type==4))
543   - {
544   - std::cout << " vehicle_pendant_det_res info:"<< endl;
545   - int vpd_num = result[b].info[c].vehicle_pendant_det_res.count;
546   - //std::cout << vpd_num<< std::endl;
547   - std::cout << " vpd_num:"<<vpd_num<< endl;
548   - for(int p=0; p<vpd_num; p++)
549   - {
550   - int index = result[b].info[c].vehicle_pendant_det_res.vpd_res[p].index;
551   - float detect_score = result[b].info[c].vehicle_pendant_det_res.vpd_res[p].confidence;
552   - int driver_copilot_info = result[b].info[c].vehicle_pendant_det_res.vpd_res[p].driver_copilot_info;
553   - int x1=result[b].info[c].vehicle_pendant_det_res.vpd_res[p].rect.left_;
554   - int y1=result[b].info[c].vehicle_pendant_det_res.vpd_res[p].rect.top_;
555   - int x2=result[b].info[c].vehicle_pendant_det_res.vpd_res[p].rect.left_+result[b].info[c].vehicle_pendant_det_res.vpd_res[p].rect.width_;
556   - int y2=result[b].info[c].vehicle_pendant_det_res.vpd_res[p].rect.top_+result[b].info[c].vehicle_pendant_det_res.vpd_res[p].rect.height_;
557   -
558   - //std::cout <<p<<" ["<<x1<<" "<<y1<<" "<<x2-x1<<" "<<y2-y1<<"] "<<detectScore<<" "<<index<<" "<<driver_copilot_info<< std::endl;
559   - std::cout << " vpd id:"<<p<<" index:"<<index<<",vpd_det_score:" <<detect_score <<",vpd_rect:[" <<x1<<"," <<y1<<"," <<x2-x1<<"," <<y2-y1<<"]"<<" driver_copilot_info:"<<driver_copilot_info<< std::endl;
560   -
561   -
562   - Point lt(x1, y1);
563   - Point rb(x2, y2);
564   - rectangle(cvImg, lt, rb, cv::Scalar(255, 255, 0), 4);
565   -
566   - //车属性人脸特征
567   - //f(index ==1)//人脸
568   - //{
569   - // std::cout << " vpd face feature:[";
570   - // /*for(int f=0;f<VPD_FACE_FEATURESIZE;f++)
571   - // {
572   - // float fea = result[b].info[c].vehicle_pendant_det_res.vpd_res[p].feature[f];
573   - // //std::cout << "feature "<<f<<":"<<fea<<std::endl;
574   - // std::cout <<fea<< " ";
575   - // }*/
576   - // std::cout << "]"<< std::endl;
577   - // //std::cout << "face feature:"<<result[b].info[c].vehicle_pendant_det_res.vpd_res[p].feature[0]<<" "<<result[b].info[c].vehicle_pendant_det_res.vpd_res[p].feature[1]<<" "<<result[b].info[c].vehicle_pendant_det_res.vpd_res[p].feature[2]<<std::endl;
578   - //}
579   - }
580   - }
581   -
582   - //7.VID车违规结果-----------
583   - if(param.vehicle_illegal_config==SY_CONFIG_OPEN && vehicle_type==0)
584   - {
585   - std::cout << " vehicle_illegal_det_res info:"<< endl;
586   - vid_result cur_result = result[b].info[c].vehicle_illegal_det_res;
587   - //副驾驶copilot
588   - {
589   - //有人没人判断，阈值0.6
590   - std::string str_person="UNCERTAINTY";
591   - if(cur_result.copilot.person.status==SOMEBODY)
592   - str_person ="PERSON";
593   - else if(cur_result.copilot.person.status==NOBODY)
594   - str_person ="NOPERSON";
595   - //cv::putText(batch_mat[b], str_person,
596   - // cv::Point(5, 25),
597   - // cv::FONT_HERSHEY_COMPLEX_SMALL,
598   - // 1.3,
599   - // cv::Scalar(255, 0, 0),
600   - // 2);
601   -
602   - //打电话，阈值0.9
603   - std::string str_call="UNCERTAINTY";
604   - if(cur_result.copilot.call.status==lEGAL)
605   - str_call ="lEGAL";//str_call ="NOCALL";
606   - else if(cur_result.copilot.call.status==ILLEGAL)
607   - str_call ="ILLEGAL";//str_call ="CALL";
608   - //cv::putText(batch_mat[b], str_call,
609   - // cv::Point(5, 50),
610   - // cv::FONT_HERSHEY_COMPLEX_SMALL,
611   - // 1.3,
612   - // cv::Scalar(255, 0, 0),
613   - // 2);
614   -
615   - //使用电话、阈值0.89
616   - std::string str_phone="UNCERTAINTY";
617   - if(cur_result.copilot.phone.status==lEGAL)
618   - str_phone ="lEGAL";//str_phone ="NOPHONE";
619   - else if(cur_result.copilot.phone.status==ILLEGAL)
620   - str_phone ="ILLEGAL";//str_phone ="USEPHONE";
621   - //cv::putText(batch_mat[b], str_phone,
622   - // cv::Point(5, 75),
623   - // cv::FONT_HERSHEY_COMPLEX_SMALL,
624   - // 1.3,
625   - // cv::Scalar(255, 0, 0),
626   - // 2);
627   -
628   - //系安全带，阈值0.5
629   - std::string str_belt="UNCERTAINTY";
630   - if(cur_result.copilot.belt.status==lEGAL)
631   - str_belt ="lEGAL";//str_belt ="SAFETYBELT";
632   - else if(cur_result.copilot.belt.status==ILLEGAL)
633   - str_belt ="ILLEGAL";
634   -
635   - //cv::putText(batch_mat[b], str_belt,
636   - // cv::Point(5, 75),
637   - // cv::FONT_HERSHEY_COMPLEX_SMALL,
638   - // 1.3,
639   - // cv::Scalar(255, 0, 0),
640   - // 2);
641   -
642   - //抽烟，阈值0.9
643   - std::string str_smoke="UNCERTAINTY";
644   - if(cur_result.copilot.smoke.status==ILLEGAL)
645   - str_smoke ="ILLEGAL";//str_smoke ="SMOKE";
646   - else if(cur_result.copilot.smoke.status==lEGAL)
647   - str_smoke ="lEGAL";//str_smoke ="NOSMOKE";
648   - //cv::putText(batch_mat[b], str_smoke,
649   - // cv::Point(5, 100),
650   - // cv::FONT_HERSHEY_COMPLEX_SMALL,
651   - // 1.3,
652   - // cv::Scalar(255, 0, 0),
653   - // 2);
654   -
655   - //std::cout << "copilot: "<<str_person<< " "<<str_call<< " "<<str_phone<< " "<<str_belt<< " "<<str_smoke<< std::endl;
656   - //std::cout << "copilot confidence: "<<cur_result.copilot.person.confidence<< " "<<cur_result.copilot.call.confidence<< " "<<cur_result.copilot.phone.confidence<< " "<<cur_result.copilot.belt.confidence<< " "<<cur_result.copilot.smoke.confidence<< std::endl;
657   - std::cout << " copilot: "<<"person:"<<str_person<< " call:"<<str_call<< " phone:"<<str_phone<< " belt:"<<str_belt<< " smoke:"<<str_smoke<< std::endl;
658   - std::cout << " copilot confidence: "<<"person:"<<cur_result.copilot.person.confidence<< " call:"<<cur_result.copilot.call.confidence<< " phone:"<<cur_result.copilot.phone.confidence<< " belt:"<<cur_result.copilot.belt.confidence<< " smoke:"<<cur_result.copilot.smoke.confidence<< std::endl;
659   -
660   - }
661   -
662   - //驾驶driver
663   - {
664   - //有人没人判断，阈值0.6
665   - std::string str_person="UNCERTAINTY";
666   - if(cur_result.driver.person.status==SOMEBODY)
667   - str_person ="PERSON";
668   - else if(cur_result.driver.person.status==NOBODY)
669   - str_person ="NOPERSON";
670   - //cv::putText(batch_mat[b], str_person,
671   - // cv::Point(5, 150),
672   - // cv::FONT_HERSHEY_COMPLEX_SMALL,
673   - // 1.3,
674   - // cv::Scalar(255, 0, 0),
675   - // 2);
676   -
677   - //打电话，阈值0.9
678   - std::string str_call="UNCERTAINTY";
679   - if(cur_result.driver.call.status==lEGAL)
680   - str_call ="lEGAL";//str_call ="NOCALL";
681   - else if(cur_result.driver.call.status==ILLEGAL)
682   - str_call ="ILLEGAL";//str_call ="CALL";
683   - //cv::putText(batch_mat[b], str_call,
684   - // cv::Point(5, 175),
685   - // cv::FONT_HERSHEY_COMPLEX_SMALL,
686   - // 1.3,
687   - // cv::Scalar(255, 0, 0),
688   - // 2);
689   -
690   - //使用电话，阈值0.89
691   - std::string str_phone="UNCERTAINTY";
692   - if(cur_result.driver.phone.status==lEGAL)
693   - str_phone ="lEGAL";//str_phone ="NOPHONE";
694   - else if(cur_result.driver.phone.status==ILLEGAL)
695   - str_phone ="ILLEGAL";//str_phone ="USEPHONE";
696   - //cv::putText(batch_mat[b], str_phone,
697   - // cv::Point(5, 200),
698   - // cv::FONT_HERSHEY_COMPLEX_SMALL,
699   - // 1.3,
700   - // cv::Scalar(255, 0, 0),
701   - // 2);
702   -
703   - //系安全带，阈值0.5
704   - std::string str_belt="UNCERTAINTY";
705   - if(cur_result.driver.belt.status==lEGAL)
706   - str_belt ="lEGAL";//str_belt ="SAFETYBELT";
707   - else if(cur_result.driver.belt.status==ILLEGAL)
708   - str_belt ="ILLEGAL";//str_belt ="NOSAFETYBELT";
709   -
710   - //cv::putText(batch_mat[b], str_belt,
711   - // cv::Point(5, 200),
712   - // cv::FONT_HERSHEY_COMPLEX_SMALL,
713   - // 1.3,
714   - // cv::Scalar(255, 0, 0),
715   - // 2);
716   -
717   - //抽烟，阈值0.9
718   - std::string str_smoke="UNCERTAINTY";
719   - if(cur_result.driver.smoke.status==ILLEGAL)
720   - str_smoke ="ILLEGAL";//str_smoke ="SMOKE";
721   - else if(cur_result.driver.smoke.status==lEGAL)
722   - str_smoke ="lEGAL";//str_smoke ="NOSMOKE";
723   - //cv::putText(batch_mat[b], str_smoke,
724   - // cv::Point(5, 225),
725   - // cv::FONT_HERSHEY_COMPLEX_SMALL,
726   - // 1.3,
727   - // cv::Scalar(255, 0, 0),
728   - // 2);
729   -
730   - //std::cout << "driver: "<<str_person<< " "<<str_call<< " "<<str_phone<< " "<<str_belt<< " "<<str_smoke<< std::endl;
731   - //std::cout << "driver confidence: "<<cur_result.driver.person.confidence<< " "<<cur_result.driver.call.confidence<< " "<<cur_result.driver.phone.confidence<< " "<<cur_result.driver.belt.confidence<< " "<<cur_result.driver.smoke.confidence<< std::endl;
732   - std::cout << " driver: person:"<<str_person<< " call:"<<str_call<< " phone:"<<str_phone<< " belt:"<<str_belt<< " smoke:"<<str_smoke<< std::endl;
733   - std::cout << " driver confidence: person:"<<cur_result.driver.person.confidence<< " call:"<<cur_result.driver.call.confidence<< " phone:"<<cur_result.driver.phone.confidence<< " belt:"<<cur_result.driver.belt.confidence<< " smoke:"<<cur_result.driver.smoke.confidence<< std::endl;
734   -
735   - }
736   - }
737   -
738   -
739   -
740   -
741   - //8.车特征
742   - if(param.vehicle_feature_config==SY_CONFIG_OPEN && (vehicle_type==0 || vehicle_type==1))
743   - {
744   - std::cout << " vehicle_fea_res info: "<<std::endl;
745   - std::cout << " [";
746   - for(int f=0;f<VA_FEATURESIZE;f++)
747   - {
748   - float fea=result[b].info[c].vehicle_fea_res.feature[f];
749   - //std::cout << "feature "<<f<<":"<<fea<<std::endl;
750   - std::cout <<fea<<" ";
751   - }
752   - std::cout <<"]"<<std::endl;
753   - //std::cout << "feature 0 :"<<result[b].info[c].vehicle_fea_res.feature[0]<<std::endl;
754   - //std::cout << "feature 1 :"<<result[b].info[c].vehicle_fea_res.feature[1]<<std::endl;
755   - //std::cout << "feature 2 :"<<result[b].info[c].vehicle_fea_res.feature[2]<<std::endl;
756   - //std::cout << "feature 3 :"<<result[b].info[c].vehicle_fea_res.feature[3]<<std::endl;
757   - //std::cout << "feature 4 :"<<result[b].info[c].vehicle_fea_res.feature[4]<<std::endl;
758   - //std::cout << "feature 5 :"<<result[b].info[c].vehicle_fea_res.feature[5]<<std::endl;
759   -
760   - //test
761   - //for(int f=0;f<10;f++)
762   - //{
763   - // float fea=result[b].info[c].vehicle_fea_res.feature[f];
764   - // std::cout << "feature "<<f<<":"<<fea<<std::endl;
765   - //}
766   - }
767   -
768   - //11.摩托车三轮车分析
769   - if(param.vehicle_motor_tricycle_analysis_config==SY_CONFIG_OPEN && (vehicle_type==2 || vehicle_type==3))
770   - {
771   - //std::cout << "mta output----"<< std::endl;
772   -
773   - if(vehicle_type==2)//摩托车
774   - {
775   - //摩托车载人
776   - //int status=result[b].info[c].mta_res.motor_manned.status;
777   - //float score=result[b].info[c].mta_res.motor_manned.confidence;
778   - //if(status == MOTOR_MANNED)
779   - // std::cout << "motor info: MOTOR_MANNED, prob: " << score << std::endl;
780   - //else if(status == MOTOR_NOT_MANNED)
781   - // std::cout << "motor info: MOTOR_NOT_MANNED, prob: " << score << std::endl;
782   -
783   - //摩托车驾驶人是否戴安全帽
784   - int status=result[b].info[c].mta_res.motor_driver_helmeted.status;
785   - float score=result[b].info[c].mta_res.motor_driver_helmeted.confidence;
786   - if(status == MOTOR_DRIVER_HELMETED)
787   - std::cout << " motor info: MOTOR_DRIVER_HELMETED, prob: " << score << std::endl;
788   - else if(status == MOTOR_DRIVER_NOT_HELMETED)
789   - std::cout << " motor info: MOTOR_DRIVER_NOT_HELMETED, prob: " << score << std::endl;
790   -
791   - }
792   - }
793   - //载人输出
794   - //if(param.vehicle_manned_config==SY_CONFIG_OPEN && (vehicle_type==0 || vehicle_type==1 || vehicle_type==3))
795   - if(param.vehicle_manned_config==SY_CONFIG_OPEN && vehicle_type!=5)
796   - {
797   - std::cout << "----------------------manned output----"<< std::endl;
798   - int hs_count = result[b].info[c].manned_res.hs_count;
799   - std::cout <<"manned_hs_count:"<< hs_count << std::endl;
800   - for(int lo=0;lo<hs_count;lo++)
801   - {
802   - int x1=result[b].info[c].manned_res.hs_rect[lo].rect.left_;
803   - int y1=result[b].info[c].manned_res.hs_rect[lo].rect.top_;
804   - int x2=result[b].info[c].manned_res.hs_rect[lo].rect.left_+result[b].info[c].manned_res.hs_rect[lo].rect.width_;
805   - int y2=result[b].info[c].manned_res.hs_rect[lo].rect.top_+result[b].info[c].manned_res.hs_rect[lo].rect.height_;
806   -
807   - std::cout <<"manned_hs_rect:"<< x1 <<" "<< y1<<" " <<x2<<" " << y2<< std::endl;
808   -
809   - Point lt(x1, y1);
810   - Point rb(x2, y2);
811   - rectangle(cvImg, lt, rb, cv::Scalar(0, 255, 255), 4);
812   - }
813   - }
814   - }
815   -
816   - //测试相似度比对函数，batch_size设为2
817   - //double similarity=va_compute_similarity(result[0].info[0].vehicle_fea_res.feature, result[1].info[0].vehicle_fea_res.feature, FEATURESIZE);
818   - //std::cout << "similarity: "<<similarity<<std::endl;
819   -
820   -
821   -
822   - //cv::Mat showImg;
823   - //showImg=batch_mat[b].clone();
824   - //cv::resize(batch_mat[b], showImg, cv::Size(batch_mat[b].cols/2,batch_mat[b].rows/2));
825   - //cv::imshow("image", showImg);
826   - //cv::waitKey(0);
827   - cv::imwrite((saveimagepath+file).c_str(),cvImg);
828   -
829   - }
830   -
831   -
832   -
833   -
834   -
835   -
836   - //delete result
837   - for(int b=0;b<batch_size;b++)
838   - {
839   - for(int c=0; c<100; c++)
840   - {
841   - if(result[b].info[c].vehicle_pendant_det_res.vpd_res!=NULL)
842   - delete[] result[b].info[c].vehicle_pendant_det_res.vpd_res;
843   -
844   - }
845   - if(result[b].info!=NULL)
846   - delete[] result[b].info;
847   - }
848   - if(result!=NULL)
849   - delete[] result;
850   -
851   - }
852   -}
853   -
854   - std::cout << "va_release---------------------"<< std::endl;
855   - va_release(&handle);
856   - //va_release(&handle111);
857   - va_acl_release();
858   -
859   - printf("va_teat info:va_batch time: t3=%.2f\n", t3/count);
860   -
861   - //oFile.close();
862   -
863   - return 0;
864   -}
src/demo/main.cpp
  View file @581a68a
... ... @@ -10,8 +10,7 @@ int main() {
10 10  
11 11 VillageParam param;
12 12 param.dev_id = 0;
13   - param.db_path = "";
14   - param.model_path = "";
  13 + param.sdk_path = ".";
15 14 int ret = village_pic_init(&vaHandle, param);
16 15  
17 16 vector<string> vec_path;
... ...
src/village_inc.h
  View file @581a68a
... ... @@ -5,10 +5,13 @@
5 5 #include <vector>
6 6 #include "vehicle_analysis.h"
7 7  
  8 +#include "opencv2/opencv.hpp"
  9 +#include "opencv2/imgcodecs/legacy/constants_c.h"
  10 +#include "opencv2/imgproc/types_c.h"
  11 +
8 12 struct VillageParam {
9 13 int dev_id;
10   - std::string db_path;
11   - std::string model_path;
  14 + std::string sdk_path;
12 15 };
13 16  
14 17 typedef struct pendant_info
... ... @@ -57,12 +60,12 @@ typedef struct VehicleInfo {
57 60  
58 61 struct LineInfo{
59 62 int line_type;
60   - std::vector<sy_point> vec_pt;
  63 + std::vector<cv::Point> vec_pt;
61 64 };
62 65  
63 66 struct SegInfo {
64 67 int seg_type;
65   - std::vector<sy_point> vec_pt;
  68 + std::vector<cv::Point> vec_pt;
66 69 };
67 70  
68 71 typedef struct AnalysisResult {
... ...