I have developed a ROS2 package for performing object detection (Microsoft COCO objects, eg car/bicycle/person etc). It listens to topic /image and publishes ROS2 Detection2DArray messages on topic /detected_objects. It uses PyTorch MobileNet to perform the object detection.
There are certainly more sophisticated vision perception systems out there, eg NVIDIA Isaac, but the aim of this package is to be easy to setup and use and also to use standard ROS2 messages.
I hope this may be of interest.
Repo available at: ROS2 COCO detector