Hi.
I set ‘TEST.RPN_POST_NMS_TOP_N=100 andTEST.RPN_NMSTHRESH=0.85’ and after completing the training, I obtained 100 features and proposals for the training, which were saved in output/rpn/default/eval/epoch In 200/val/detections/data/. when more than one object is detected, how should we distinguish which different objects belong to each of the 100 proposals?
For example, if there are 5 cars in the detection result, how can we distinguish which vehicles belong to the first, second, third, fourth, and fifth of the generated 100 proposals?
Thanks a lot.
Hi. I set ‘TEST.RPN_POST_NMS_TOP_N=100 andTEST.RPN_NMSTHRESH=0.85’ and after completing the training, I obtained 100 features and proposals for the training, which were saved in output/rpn/default/eval/epoch In 200/val/detections/data/. when more than one object is detected, how should we distinguish which different objects belong to each of the 100 proposals? For example, if there are 5 cars in the detection result, how can we distinguish which vehicles belong to the first, second, third, fourth, and fifth of the generated 100 proposals? Thanks a lot.