tensorrtx
tensorrtx copied to clipboard
Integration of this project in a ROS code
Has anyone tried to use this project as a ROS node? I am currently trying one, but facing errors while compiling the whole project with problems related to gcc version.
Yes. I was using Melodic on Jetson NX with JP 4.6 I was only building the Yolov5 sub director. I made a copy of the main.cpp file and only built the "inference side" There is a simple topic for subscribing and one for publishing. I did not make use of the batch processing but given more thought it would be cool to have it; aggregate many input streams and do a single forward pass.
@jwrl7 Can you please share with me your project(also the ros-part) because I'm trying to do the same but I have multiple errors there and probably I'm missing something there. Thank you very much in advance
Especially I'm interested in how do you use the main function in yolov5.cpp file. Do you use the main function to be a callback for an image ? I've tried to create a subscriber for the camera images and used the main function in yolov5.cpp(which I renamed to objectDetectionCallback) to be a callback for the image topic, but for now I get so many different errors....
@jwrl7 I'm trying to integrate to ROS. Unlike yours, I am building all the code and whenever I run my ROS node, it is building engine again and again. Which part of code should be built for get rid of "building engine loop"?
This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.