Why does Model Optimizer require length of shapes of arrays from model to be 1?

[K__Mike]

Hello,

When optimizing tensorflow model, I had 4 nodes Fatten_1

flatten_1/stack/0               flatten_1/stack/0/Output_0/Data_190

flatten_1/stack                 flatten_1/stack/Output_0/Data_248

2nd half causes optimization fail as value/shape propagation stops.

Is VINO not compatible with all model variations in AI?

Also posted here -

https://stackoverflow.com/questions/55420326/cannot-infer-shapes-or-values-for-node-flatten-1-stack

Please illuminate @Subha.

[Shubha_R_Intel]

Dear Mike, it's difficult to say what happened based on that Stack Overflow link and your post as well. Can you kindly re-run Model Optimizer with --log_level DEBUG and copy the ERROR part here ?

Thanks,

Shubha