I have trained a tensorflow object detection model which will detect a box, bag and crate. when i tested that trained model it is giving more false positives. I trained that with the cleaned datas with proper annotations, though it is giving more false positives. Is there any way to reduce the false positives?
I am using ssd_mobilenet_v1_coco to train a model. Do i need to change anything in the pipeline.config file to reduce the false positives?