high-filter

src/camera_bev_infer/configs/CameraBevParam.yaml

 rgb_sub_topic_name: /sensor/camera/front_middle/image/bgr
 pointcloud_sub_topic_name: /total_calib_pointcloud
-detection_pub_topic_name: /perception/camera/obstacle
+detection_pub_topic_name: /camera/objects
 # static_gridmap_pub_topic_name: /perception/camera/bev_static_gridmap
 creator_id: 0000
 creator_name: camera_bev_node
@@ -15,7 +15,7 @@ vis_frame_id: Lidar
 vis_project: true
 # ["pedestrian", "vehicle", "stone"]
 pub_classes: ["stone", "cone"]
-obstacle_height_threshold: 5 # cm
+obstacle_height_threshold: 25 # cm
 camera_intrinsic: [1346.61241105747, 0, 951.014999441565,
                     0, 1344.43795511862, 530.562496955385,
                     0, 0, 1]

src/camera_bev_infer/configs/CameraBevParam_jd06.yaml

 rgb_sub_topic_name: /front_rgb/image
 pointcloud_sub_topic_name: /total_calib_pointcloud
-detection_pub_topic_name: /perception/camera/obstacle
+detection_pub_topic_name: /camera/objects
 # static_gridmap_pub_topic_name: /perception/camera/bev_static_gridmap
 creator_id: 0000
 creator_name: camera_bev_node

src/camera_bev_infer/src/vision_transform/vision_3d.cpp

@@ -5,7 +5,7 @@
  * @Date: 2023-09-03 03:13:41
  * @email: xin.wang@waytous.com
  * @LastEditors: wxin
- * @LastEditTime: 2024-02-28 10:32:38
+ * @LastEditTime: 2024-03-12 06:38:25
  */
 
 
@@ -1024,7 +1024,7 @@ int Visione_3D::Process_Instance_Objects(object_Seg_info& obj, const cv::Mat& in
                 }
             }
 
-            bool filtered = (max_z - min_z < obstacle_height_threshold * 0.01) && (max_z > obstacle_height_threshold * 0.01 * 0.5);
+            bool filtered = (max_z - min_z < obstacle_height_threshold * 0.01) || (max_z < obstacle_height_threshold * 0.01 * 0.5);
             if(!filtered){
                 Tracks[obj.track_id].valid_frame_num++; // add
                 std::cout << "------------------------ "<< obj.track_id <<": " << obj.name <<" with "<<points3d.size() << 
@@ -1050,9 +1050,6 @@ int Visione_3D::Process_Instance_Objects(object_Seg_info& obj, const cv::Mat& in
 
 // 同时使用深度估计和点云投影来获取bev box
 int Visione_3D::Process_objects(const cv::Mat& image, const cv::Mat& depth, Cloud::Ptr& pcl_cloud, std::vector<object_Seg_info>&objs, BEV_GridMap& vGrid){
-    if(depth.empty()){
-        return Process_objects(image, pcl_cloud, objs, vGrid);
-    }
     if(pcl_cloud == nullptr){
         return Process_objects(image, depth, objs, vGrid);
     }
@@ -1109,9 +1106,11 @@ int Visione_3D::Process_objects(const cv::Mat& image, const cv::Mat& depth, Clou
             int res = Process_Instance_Objects(obj, point_index_in_camera, pcl_cloud, vGrid);
             if(res == -1){ // valid points太少，使用深度估计
                 // std::cout<<"using depth " << obj.name << "," <<obj.fConfidence<< std::endl;
+                if(!depth.empty()){
                     Process_Instance_Objects(obj, depth, pTable, vGrid);
                 }
             }
+        }
         else{
             Process_Segmantic_Objects(obj, point_index_in_camera, pcl_cloud, vGrid);
         }