Hey we got a lot out of this project GitHub - NVIDIA-AI-IOT/YOLOv5-with-Isaac-ROS: Sample showing how to use YOLOv5 with Nvidia Isaac ROS DNN Inference
and we want to add a tracker in realtime on the basis of this project.
DO you have a recommendation or a built in tracker to use with the nvidia issac yolov5?
Once you have bounding boxes from a detector such as YOLOv5 or DetectNet, you can use a tracker such as the one from VPI (KLT bounding box tracker) that can use the image features in the bounding box and follow them through optical flow frame-to-frame. When you run detection on a given frame, you can use the tracker to follow the object in realtime and to perform correspondence matching in the next frame you run detection on (e.g., “box 7 in frame 1 is the same as box 9 in frame 0”).