运行yolo得到engine文件
需要制作权重文件kuangka.wts,然后得到engine文件
n_camera_obj节点
hrnet_rt模型
detect和seg模型一起跑
roadside激光相机obj节点
# lidar obj
rosrun n_point_cloud_segmentation n_point_cloud_segmentation
# camera obj
rosrun n_camera_hive_detect_obj n_camera_hive_detect_obj src/nodes/n_camera_hive_detect_obj/src/roadside_det_1022.engine /miivii_gmsl_ros/n_camera_miivii_gmsl_prep n_camera_hive_detect_obj
rosrun n_camera_hive_detect_obj n_camera_hive_detect_obj src/engine/roadside.engine.11.01 /miivii_gmsl_ros/n_camera_miivii_gmsl_prep
标定数据录制
相机节点原始图像开启(Roadside_Env_Detect):
rosrun n_camera_miivii_gmsl_camera_prep n_camera_miivii_gmsl_camera_prep
激光标定到地面的点云:
1.驱动
rosrun rslidar_sdk rslidar_sdk_node
2.检测
rosrun n_point_cloud_segmentation n_point_cloud_segmentation
录制标定原素材:
rosbag record /miivii_gmsl_ros/n_camera_miivii_gmsl_prep /rslidar_points
标定程序录制数据
- 图像解析每一个相机(Roadside_Camera_Parse):
rosrun n_camera_miivii_gmsl_camera_prep n_camera_visualization 0 /miivii_gmsl_ros/n_camera_miivii_gmsl_prep
- 录制激光雷达标定到0平面的数据(Lidar_Base_Calibration):
修改参数,获得0平面点云
3.录制0平面激光点云与某一个相机的解析图像
rosbag record /n_camera_0_show /cal_lidar_points
rosrun n_camera_miivii_gmsl_camera_prep n_camera_visualization 1 /miivii_gmsl_ros/n_camera_miivii_gmsl_prep
rosbag record /n_camera_1_show /cal_lidar_points
4.运行(Lidar_Calibration节点)订阅 /n_camera_1_show /cal_lidar_points
rosrun CLCalibrator CLCalibrator_node
sudo rename 's/:/-/' *
启动相机节点
// detect
rosrun n_camera_hive_detect_obj n_camera_hive_detect_obj src/engine/roadside.engine.12.03 /miivii_gmsl_ros/n_camera_miivii_gmsl_prep
// project
rosrun n_camera_hive_detect_obj n_camera_project /n_camera_hive_detect
// track
rosrun n_camera_track n_camera_track /n_camera_project
// fit
rosrun n_camera_fit n_camera_fit /n_camera_track