I am working on a task where I have to detect people and do activity recognition. I am using **Detectron2 **Keypoint detection model.There are many False negatives with the detection. So my solution to reduce the FN's is to train the custom detection part of Detectron2 . I trained the custom detection model and got the improved results. But now I have to change the backbone of the Detectron2 object detection part with my custom detection model.
I changed the config file but that didn't work.
