OpenVINO Internal Error, shapes not aligned

[Ravindran__Isaac]

Hi guys,

I was attempting to prepare my model to be uploaded to the NCS2. However, I kept getting this error when running the model optimizer:

[ ERROR ]  -------------------------------------------------
[ ERROR ]  ----------------- INTERNAL ERROR ----------------
[ ERROR ]  Unexpected exception happened.
[ ERROR ]  Please contact Model Optimizer developers and forward the following information:
[ ERROR ]  shapes (1024,10) and (0,) not aligned: 10 (dim 1) != 0 (dim 0)
[ ERROR ]  Traceback (most recent call last):
  File "/opt/intel/openvino_2019.1.144/deployment_tools/model_optimizer/mo/main.py", line 312, in main
    return driver(argv)
  File "/opt/intel/openvino_2019.1.144/deployment_tools/model_optimizer/mo/main.py", line 263, in driver
    is_binary=not argv.input_model_is_text)
  File "/opt/intel/openvino_2019.1.144/deployment_tools/model_optimizer/mo/pipeline/tf.py", line 153, in tf2nx
    for_graph_and_each_sub_graph_recursively(graph, fuse_linear_ops)
  File "/opt/intel/openvino_2019.1.144/deployment_tools/model_optimizer/mo/middle/pattern_match.py", line 58, in for_graph_and_each_sub_graph_recursively
    func(graph)
  File "/opt/intel/openvino_2019.1.144/deployment_tools/model_optimizer/mo/middle/passes/fusing/fuse_linear_ops.py", line 275, in fuse_linear_ops
    is_fused = _fuse_add(graph, node, fuse_nodes, False)
  File "/opt/intel/openvino_2019.1.144/deployment_tools/model_optimizer/mo/middle/passes/fusing/fuse_linear_ops.py", line 209, in _fuse_add
    fuse_node.in_node(2).value += np.dot(fuse_node.in_node(1).value, value)
ValueError: shapes (1024,10) and (0,) not aligned: 10 (dim 1) != 0 (dim 0)

[ ERROR ]  ---------------- END OF BUG REPORT --------------
[ ERROR ]  -------------------------------------------------

Any help would be greatly appreciated!

Thanks

Isaac

P.S. Please find attached the model PB file

[Shubha_R_Intel]

Dear Ravindran, Isaac,

What is your model optimizer command ?

Thanks,

Shubha

[Ravindran__Isaac]

Hi Shubha,

I used the following command:

python /opt/intel/openvino_2019.1.144/deployment_tools/model_optimizer/mo_tf.py --input_model frozen_colour_model.pb --output_dir . --input_shape [1,96,96,3] --data_type FP32

Thanks

Isaac

[Shubha_R_Intel]

Dear Ravindran, Isaac,

I also got an error but different from you, as I'm using the latest version of OpenVino 2019R1.1. Also for NCS2 you should be using --data_type FP16 not FP32 but that doesn't impact the Model Optimizer error you're getting.

May I ask from where did you get this model ? Model Optimizer Tensorflow Supported List contains the list of supported Tensorflow Models.

Please report back as to where you got your model,

Thanks,

Shubha