Error when converting mxnet to pb

[Zheweiqiu]

Platform (like ubuntu 16.04/win10): ubuntu 16.04 Python version: 3.7.3 Source framework with version (like Tensorflow 1.4.1 with GPU): mxnet 1.5.0 Destination framework with version (like CNTK 2.3 with GPU): pb Pre-trained model path (webpath or webdisk path): Pre-trained model available from here under "RetinaFace Pretrained Models" Running scripts: python -m mmdnn.conversion._script.convertToIR -f mxnet -n R50-symbol.json -w R50-0000.params -d resnet50 --inputShape 3,640,640 Got an warning: UserWarning: You created Module with Module(..., label_names=['softmax_label']) but input with name 'softmax_label' is not found in symbol.list_arguments(). Did you mean one of: data Size of three files I got: 20K resnet50.json 496K resnet50.npy 4.0K resnet50.pb

Have no idea how to fix this. Any help will be appreciate!

[AlonSh]

I'm having the same issue. The original module has many more layers and weights than the generated module. @Zheweiqiu did you find a solution?

[cydawn]

I am converting pre-trained models of RetinaFace implemented by mxnet. I have the same problem. Did you fix this problem? @Zheweiqiu @AlonSh

[AlonSh]

@cydawn I found a pytorch implementation, rewrote the model in keras and found a tool that loads pytorch params into tf graphs. I'm sorry but that work was done in my job using internal tools so I cannot share it /: Good luck mate