wkentaro/fcn

How to get training feature maps?

dxiaosa opened this issue · 1 comments

Last time, we resolved the problem of fetching the feature maps off line, ie. in testing program. So, if I want to get the feature maps when training, how can I do? Thx.

Do you mean the feature maps are the outputs of each hidden layer?
If so, you can get them by setting self.conv1_1 = h here: https://github.com/wkentaro/fcn/blob/master/fcn/models/fcn32s.py#L54
and self.model.conv1_1 in the test program.