Closed acecreamu closed 6 years ago
Using train.py as example, you can run sess.eval on raw_output. This would give a downsample score map, so if you want the full size one, you will need to do interpolation, too
I got good result just modifying inference.py. Thanks for answer, anyway.
Hi Vladimir,
Could you give a pice of advice, how can I extract the score map (input before softmax function) from pretrained model?
Thank you in advance