Boost.Python.ArgumentError

[yorkjohn]

Platform (like ubuntu 16.04/win10): ubuntu 16.04

Python version:2.7

Source framework with version (like Tensorflow 1.4.1 with GPU):Tensorflow with GPU

Destination framework with version (like CNTK 2.3 with GPU):caffe with GPU

transfer model from TF to caffe has this error.
This issue has been aroused in the closed topic " Trouble Converting IR to Caffe #111 "
it can be solved using str() method.
However, when converting straight from TF to caffe, same problem occurs when the inter-media py file is
created. For example: mmconvert -sf tensorflow -in vgg16.ckpt.meta -iw vgg16.ckpt -df caffe --dstNodeName MMdnn_Output -om ./vgg16_TF2caffe

[JiahaoYao]

Platform : MacOS 10.13.3

Python version:2.7

@yorkjohn
I think it can be still solved by using str() method.

In mmdnn/conversion/caffe/saver.py, I add in Line 7-8

    dump_net = str(dump_net)
    dump_weight = str(dump_weight)

The code crashes because in model = caffe.Net(dump_net, dump_weight, caffe.TEST), dump_net and dump_weight are 'unicode' not 'str'. It works after converted to str. This problem is also mentioned in Caffe Issue #3220.

I will provide pull request soon to solve this problem.