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.
6392 Discussions

Problem converting a vggface based keras model to IR format

Cañas__Jhovanny
Beginner
374 Views

I try to convert my vggface based model into keras with backend tensorflow to IR format t using mo.py in Openvino, My model is frozen. I attach the log

Thank!.

 

0 Kudos
1 Reply
Shubha_R_Intel
Employee
374 Views

Dear Cañas, Jhovanny,

Would you mind chopping everything after training/RMSprop/Variable_7 off ? Or everything after dense_4/kernel off ? If these happen toward the end of the model, maybe it's OK. The technique I'm suggesting is called Model Cutting . You can add --output "dense_4/kernel" or --output "training/RMSprop/Variable_7" to your mo_tf.py command which creates a new exit point for the model. This may be OK if these occur toward the end of the model and there's nothing important (such as an output layer) afterward.

Please let me know the answer on this forum.  The exception coming from Model Optimizer is clear :

Graph contains 0 node after executing <class 'extensions.front.input_cut.InputCut'>. It considered as error because resulting IR will be empty which is not usual

There is a sub-graph in your model which contains zero nodes. Maybe chopping it off will solve the problem.

Thanks,

Shubha

 

0 Kudos
Reply