openvino conversion model error

[zuhair__mohamed]

  Hi,

We have installed and tried some sample demo applications and its working fine. But when we tried with our own in-build model we faced some issues. pls help  to solve that.  i have attached the screenshots also.

issue:   " Conversion of the model(maybe due to variable size input of model)(in case of object_Detection) is creating the issue. Even it is created, then the model is not able to load correctly.  If the model is loaded correctly(in case of yolo_v3), then ie_api.pyx error is shown. "

The original model of Object Detection is converted to the Inference Engine format. But the issue of ie_api.pyx is still there. we have tried ubuntu also but same issue.

Can you please suggest on this. Also please find the attachment openvino.rar.

Best Regards,

Zuhair

[Cary_P_Intel1]

Hello, Mohamed,

From your screen shot of error message, I assume you use the faster-rcnn which has the proposal layer, and the proposal layer support requires you to point to the extension. Please refer to the ssd python sample under the "<OpenVINO_root>/deployment_tools/inference_engine/samples/python_samples/object_detection_sample_ssd/object_detection_sample_ssd.py" to understand how to use the latest "IECore" class and linking to the extension file.