OpenVINO Internal Error

[Byung-Hun__Park]

Hi, I am testing my DNN model on linux using version 2018.3.343 of OpenVINO. I have installed OpenVINO correctly refering to the OpenVINO Installation Guide. But while running model-optimizer, internal error occurs and the log is following. On previous version 2018.2.319, always succeeded.

Model Optimizer arguments:
Common parameters:
 - Path to the Input Model:  /home/dna01/Projects/Evo/Workspace/CaffeModels/LPBQD/train1-ss_iter_33500.caffemodel
 - Path for generated IR:  /home/dna01/Projects/Evo/Workspace/CaffeModels/LPBQD/FP16
 - IR output name:  LPBQD-Extra
 - Log level:  ERROR
 - Batch:  1
 - Input layers:  conv-qloc_c1,conv-qloc_c2,conv-qloc_c3,conv-qloc_c4
 - Output layers:  sm-qloc_c1,sm-qloc_c2,sm-qloc_c3,sm-qloc_c4
 - Input shapes:  Not specified, inherited from the model
 - Mean values:  Not specified
 - Scale values:  Not specified
 - Scale factor:  Not specified
 - Precision of IR:  FP16
 - Enable fusing:  True
 - Enable grouped convolutions fusing:  True
 - Move mean values to preprocess section:  False
 - Reverse input channels:  False
Caffe specific parameters:
 - Enable resnet optimization:  True
 - Path to the Input prototxt:  /home/dna01/Projects/Evo/Workspace/CaffeModels/LPBQD/deploy.prototxt
 - Path to CustomLayersMapping.xml:  Default
 - Path to a mean file:  Not specified
 - Offsets for a mean file:  Not specified
Model Optimizer version:  1.2.185.5335e231
[ ERROR ]  -------------------------------------------------
[ ERROR ]  ----------------- INTERNAL ERROR ----------------
[ ERROR ]  Unexpected exception happened.
[ ERROR ]  Please contact Model Optimizer developers and forward the following information:
[ ERROR ]  '<' not supported between instances of 'NoneType' and 'int'
[ ERROR ]  Traceback (most recent call last):
  File "/home/dna01/local/opt/intel/computer_vision_sdk_2018.3.343/deployment_tools/model_optimizer/mo/main.py", line 321, in main
    return driver(argv)
  File "/home/dna01/local/opt/intel/computer_vision_sdk_2018.3.343/deployment_tools/model_optimizer/mo/main.py", line 275, in driver
    custom_layers_mapping_path=custom_layers_mapping_path)
  File "/home/dna01/local/opt/intel/computer_vision_sdk_2018.3.343/deployment_tools/model_optimizer/mo/pipeline/caffe.py", line 231, in driver
    input_names=input_names)
  File "/home/dna01/local/opt/intel/computer_vision_sdk_2018.3.343/deployment_tools/model_optimizer/mo/pipeline/common.py", line 76, in prepare_emit_ir
    port_renumber(graph)
  File "/home/dna01/local/opt/intel/computer_vision_sdk_2018.3.343/deployment_tools/model_optimizer/mo/back/ie_ir_ver_2/emitter.py", line 335, in port_renumber
    for u, d in get_sorted_inputs(node):
  File "/home/dna01/local/opt/intel/computer_vision_sdk_2018.3.343/deployment_tools/model_optimizer/mo/graph/graph.py", line 290, in get_sorted_inputs
    return sorted([x for x in get_inputs(node.graph, node.node) if 'in' in x[1]], key=lambda x: x[1]['in'])
TypeError: '<' not supported between instances of 'NoneType' and 'int'

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

[Monique_J_Intel]

Hi,

Do you mind attaching your model's files so I can reproduce the issue on my end and providing the command line you used?

Kind Regards,

Monique Jones

[Nofil_F_Intel]

I have a similar issue when I am trying to run a Mxnet (Resnet50 model). 

Here is the bug report.

Model Optimizer version: unknown version

[ ERROR ]  -------------------------------------------------

[ ERROR ]  ----------------- INTERNAL ERROR ----------------

[ ERROR ]  Unexpected exception happened.

[ ERROR ]  Please contact Model Optimizer developers and forward the following information:

[ ERROR ]  3

[ ERROR ]  Traceback (most recent call last):

  File "/home/nfawad/openvino/dldt/model-optimizer/mo/main.py", line 312, in main

    return driver(argv)

  File "/home/nfawad/openvino/dldt/model-optimizer/mo/main.py", line 278, in driver

    ret_res = mo_mxnet.driver(argv, argv.input_model, model_name, argv.output_dir)

  File "/home/nfawad/openvino/dldt/model-optimizer/mo/pipeline/mx.py", line 95, in driver

    convert_batch_norm(graph)

  File "/home/nfawad/openvino/dldt/model-optimizer/mo/middle/passes/fusing/decomposition.py", line 48, in convert_batch_norm

    mean = node.in_node(3)

  File "/home/nfawad/openvino/dldt/model-optimizer/mo/graph/graph.py", line 139, in in_node

    return self.in_nodes(control_flow=control_flow)[key]

KeyError: 3

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

[ ERROR ]  -------------------------------------------------

Command line used:

python3 mo.py --framework mxnet --input_model mxnetmodels/resnet50/resnet-50-0000.params --input_shape [1,3,224,224]

[Shubha_R_Intel]

Dear Nofil F.,

From where are you getting this MxNet model ?  Are you following The MO MxNet Conversion Doc instructions ?

Thanks,

Shubha