Intel® Distribution of OpenVINO™ Toolkit
Community assistance about the Intel® Distribution of OpenVINO™ toolkit, OpenCV, and all aspects of computer vision-related on Intel® platforms.

Unable to use the SRGAN Super Resolution model

Niranjan__Udipi
Beginner
383 Views

I tried to convert the SRGAN model available from https://github.com/WolframRhodium/Super-Resolution-Zoo/tree/master/SRGAN/SRResNet%40PyTorch using the following model converter command:

 ./mo_mxnet.py --input_model ~/Downloads/Zoom-Experiments/Openvino-SR/SRResNet_4x-0000.params  --output_dir /tmp/ --input_shape [1,3,96,96]

I get the following error message:

Model Optimizer version: 2019.3.0-408-gac8584cb7
[ ERROR ]  The attribute "shape" of the operation "reshape0" contains value "-4" which is not supported.
[ ERROR ]  Model Optimizer does not support layer "reshape0". Please, implement extension.

Where are the instructions on what parameters to be used and how to port the models in https://github.com/WolframRhodium/Super-Resolution-Zoo, to the OpenVino tool?

Thanks in anticipation,

Niranjan

 

0 Kudos
1 Reply
JesusE_Intel
Moderator
383 Views

Hi Niranjan,

Thanks for reaching out. The model you are trying to use has layers that are not supported by the OpenVINO toolkit. You may be able to implement custom extensions for these layers. Take a look at the Custom Layers Guide for additional information.

Hope this helps.

Regards,

Jesus

0 Kudos
Reply