When I used ROS1/Noetic, I found quite a lot of well developed packages for vision. I have not had that much success finding packages for Humble. I have only just started using ROS2, so maybe I am missing something.
I'd like a package for face recognition and another for object recognition (probably yolo-based). Any recommendations?
I found, and tested this package for visual object recognition. It works fine using Ubuntu 22.04/ROS2 Humble apart from excessive GPU usage when using a live video stream. That is a long standing problem, to which I discovered a hacky fix in 2019. I had to update the fix for ros2 (due to 'seq' being removed from std_msgs/Header). The full code for the fix is given in this message.
I have not found any ros2 packages for face recognition, and I have had no suggestions from anybody, so I will assume that none exist.
I have tried to put together a face recognition package for ros2/Humble. It is in my repositories face_recog and vision_interfaces
This is the first time that I have tried to make a publicly available ros package, and I have had zero support, so it is probably not done well. If I get any suggestions for improvements, I will try to incorporate them.
