TypeError: must be str, not bytes

[sdberardinelli]

When running dump_filters:

python dump_filters.py 
>> Downloading inception-2015-12-05.tgz 100.0%
Succesfully downloaded inception-2015-12-05.tgz 88931400 bytes.
W tensorflow/stream_executor/cuda/cuda_driver.cc:590] creating context when one is currently active; existing: 0x5f52e40
W tensorflow/stream_executor/cuda/cuda_driver.cc:590] creating context when one is currently active; existing: 0x2ee8240
W tensorflow/stream_executor/cuda/cuda_driver.cc:590] creating context when one is currently active; existing: 0x2eea560
W tensorflow/core/framework/op_def_util.cc:332] Op BatchNormWithGlobalNormalization is deprecated. It will cease to work in GraphDef version 9. Use tf.nn.batch_normalization().
169 0.891073
75 0.00779056
7 0.00295913
325 0.00146577
878 0.00117424
Traceback (most recent call last):
  File "dump_filters.py", line 292, in <module>
    tf.app.run()
  File "/usr/lib/python3.6/site-packages/tensorflow/python/platform/app.py", line 44, in run
    _sys.exit(main(_sys.argv[:1] + flags_passthrough))
  File "dump_filters.py", line 289, in main
    run_inference_on_image(image)
  File "dump_filters.py", line 137, in run_inference_on_image
    dump_convbn(sess, 'conv')
  File "dump_filters.py", line 69, in dump_convbn
    padding = make_padding(conv.get_attr("padding"), weights.shape)
  File "dump_filters.py", line 62, in make_padding
    sys.exit('Invalid padding name '+padding_name)
TypeError: must be str, not bytes

I'm not sure why this happens, but I was able to work around this by adding

padding_name = (padding_name.decode("utf-8"))

in make_padding before the conditions. After adding this it seems to work fine.