When I use the YOLO_weight_extractor why the max-pool layer are not generated ?

Question

When I use the YOLO_weight_extractor why the max-pool layer are not generated ?

Closed this issue 8 years ago · 2 comments

Answer 1 · 2017-04-20T01:57:06.000Z

Hi! weight extractor extracts values of weights and biases. Max-Pool layers don't have any weights or biases so we don't need to generate pooling layers in weight extractor. When the YOLO.py builds the tensorflow network, it builds pooling layers correctly:)

Answer 2 · 2017-04-20T02:28:59.000Z

Thanks a lot ! I ran the YOLO_builder.py successfully just now, but the generated files are .ckpt.data file and .ckpt.meta. Could you please tell me why two files generated and which one can be used ? Thank you~