Use the Intel D435 real-sensing camera to realize object detection based on the Yolov3-5 framework under the Opencv DNN(old version)/TersorRT(now) by ROS-melodic.Real-time display of the Pointcloud in the camera coordinate system.
When I try to compile it results in a error saying:
/home/abc/catkin_ws/src/realsense_d435/src/Subscriber_realsense.cpp:8:10: fatal error: realsense_d435/objectionsofonemat.h: No such file or directory
#include "realsense_d435/objectionsofonemat.h"
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
compilation terminated.
Thank you for your repo! Just some quick notes which weren't immediately clear:
this repo requires librealsense >= 2.34.0 (and not >=2.0 as per readme) otherwise method rs2::depth_frame::get_units() doesn't exist;
there is a tiny mistake with the initialization of array<int,3> Center_Point{0,0,0}; which should (likely) be array<int,3> Center_Point{{0,0,0}}; with double braces, if that's a std::array;
the realsense_d435 package is missing message dependencies for its own messages in the publisher/subscriber targets, which makes compilation not consistent as sometimes the header objection.h doesn't get generated in time for compilation;
the package is also missing a dependency on realsense2_camera, specifically build+exec depend in the package.xml and a CMakeLists find_package dependency as to build the publisher there's a need for Extrinsics.h header from that package's messages.
After these it seems to build fine except for my workstation currently missing CUDA. You actually mention only having to need OpenCV DNN as the "old" way of making it work. What changes would be required for this?
Hi @Mazhichaoruya Thanks a lot for your repository.
I'm Trying Object detection using RealsenseL515 on CPU. While creating(second time) catkin_work I got error
Can you please help me to resolve this issue and required changes that I need to commit in the files
(Since I'm a Undergraduate student I don't have that much idea)