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.

Segmentation issue with R12019

Rushikesh_P_Intel
352 Views

After the release of openVINO 2019R1 version customer is facing two additional issues.

  1. Segmentation fault issue during infer request, while running the algorithm on CPU. This issue is not there in the 2018R5 version. getting stuck at point 5 in below image:
  2. Additionally, as per the openVINO 2019R1 release notes it says that there is added support for Expand layer,  But again the model optimizer fails to convert the design to xml and bin files. For your reference they are using ONNX Expand layer. Please help to confirm if we are missing something here.

    Below is the log when the Expand layer failure occurs:

     

    [ WARNING ]  Instructions/layers that do not have attribute extractors:
    [ WARNING ]      Expand (3)
    [ WARNING ]          484
    [ WARNING ]          492
    [ WARNING ]          498
    ...

    [ ERROR ]  Cannot infer shapes or values for node "484".
    [ ERROR ]  There is no registered "infer" function for node "484" with op = "Expand". Please implement this function in the extensions.
     For more information please refer to Model Optimizer FAQ (<INSTALL_DIR>/deployment_tools/documentation/docs/MO_FAQ.html), question #37.
    [ ERROR ] 
    [ ERROR ]  It can happen due to bug in custom shape infer function <UNKNOWN>.
    [ ERROR ]  Or because the node inputs have incorrect values/shapes.
    [ ERROR ]  Or because input shapes are incorrect (embedded to the model or passed via --input_shape).
    [ 2019-04-08 15:31:33,632 ] [ DEBUG ] [ infer:194 ]  Node "484" attributes: {'pb': input: "237"
    input: "483"
    output: "484"
    op_type: "Expand"
    ...
    Model Optimizer arguments:
    Common parameters:
        - Path to the Input Model:     <model>.onnx
        - Path for generated IR:     <model>/.
        - IR output name:     <model>
        - Log level:     DEBUG
        - Batch:     Not specified, inherited from the model
        - Input layers:     Not specified, inherited from the model
        - Output layers:     Not specified, inherited from the model
        - Input shapes:     Not specified, inherited from the model
        - Mean values:     Not specified
        - Scale values:     Not specified
        - Scale factor:     Not specified
        - Precision of IR:     FP32
        - Enable fusing:     True
        - Enable grouped convolutions fusing:     True
        - Move mean values to preprocess section:     False
        - Reverse input channels:     False
    ONNX specific parameters:
    Model Optimizer version:     2019.1.0-341-gc9b66a2

0 Kudos
1 Reply
Shubha_R_Intel
Employee
352 Views

Dear Purohit, Rushikesh,

Please tell me more about the segmentation fault. Exactly how does it happen ? Did the customer run one of the OpenVino samples ?  Also, which model is your customer having issues with ? Is it a private model ? You are an Intel employee so you can send me a *.zip of the model if you'd like and I will investigate it.

Thanks,

Shubha

0 Kudos
Reply