simple-faster-rcnn-pytorch
simple-faster-rcnn-pytorch copied to clipboard
Time consumption mainly comes from the i/o of cpu and gpu(such as numpy to tensor(cuda)) is that correct?