caffe-tensorflow icon indicating copy to clipboard operation
caffe-tensorflow copied to clipboard

No support for custom python layer

Open tkasarla opened this issue 8 years ago • 5 comments

I am trying to convert FCN for semantic segmentation model to tensorflow and following is the error I get:

Error encountered: Unknown layer type encountered: Python The first layer is a custom python layer defined by the network itself to read the data.

Will there be support added in future?

tkasarla avatar May 24 '17 05:05 tkasarla

Hello @tejaswikasarla, I am having the same problem, trying to convert a GVV caffe network to tf. Did you find the solution?

sthalles avatar Nov 09 '17 17:11 sthalles

Any recommendation on how to handle custom Caffe layers? I'm having the same issue with an application-specific network.

asutic avatar Apr 24 '18 05:04 asutic

I didn't find any solution for this. I hope the author of the code has a workaround to suggest.

tkasarla avatar Apr 25 '18 06:04 tkasarla

Same issue here. Hope to get an update :-(

coldmanck avatar May 23 '18 16:05 coldmanck

dealing with the custom layer https://ndres.me/post/convert-caffe-to-tensorflow/

Ai-is-light avatar Aug 24 '18 03:08 Ai-is-light