Community
cancel
Showing results for 
Search instead for 
Did you mean: 
liu__zhiqi
Beginner
136 Views

unable to convert the Caffe VGG16 model,

Jump to solution

 

C:\zhiqi_work\Intel_OpenVI\computer_vision_sdk_2018.1.265\deployment_tools\model_optimizer>python mo_caffe.py --input_model C:\zhiqi_work\Intel_OpenVI\computer_vision_sdk_2018.1.265\deployment_tools\caffe_model\VGG16_faster_rcnn_final.caffemodel  --model_name C:\zhiqi_work\Intel_OpenVI\computer_vision_sdk_2018.1.265\deployment_tools\caffe_model\deploy.prototxt
Model Optimizer arguments
        Batch:  1
        Precision of IR:        FP32
        Enable fusing:  True
        Enable gfusing:         True
        Names of input layers:  inherited from the model
        Path to the Input Model:        C:\zhiqi_work\Intel_OpenVI\computer_vision_sdk_2018.1.265\deployment_tools\caffe_model\VGG16_faster_rcnn_final.caffemodel
        Input shapes:   inherited from the model
        Log level:      ERROR
        Mean values:    ()
        IR output name:         C:\zhiqi_work\Intel_OpenVI\computer_vision_sdk_2018.1.265\deployment_tools\caffe_model\deploy.prototxt
        Names of output layers:         inherited from the model
        Path for generated IR:  C:\zhiqi_work\Intel_OpenVI\computer_vision_sdk_2018.1.265\deployment_tools\model_optimizer
        Reverse input channels:         False
        Scale factor:   None
        Scale values:   ()
        Version:        0.3.75.d6bae621
        Input proto file:       deduced from the input model
        Path to CustomLayersMapping.xml:        extensions/front/caffe/CustomLayersMapping.xml
        Path to a mean file:
        Offsets for a mean file:        None
[ ERROR ]  Found custom layer proposal. Model Optimizer does not support this layer. Please, register it in CustomLayersMapping.xml or implement extension. For more information please refer to Model Optimizer FAQ, question #45.

this above is the output from the command. I followed the instructions from the sample site, but why it doesn't work?

 python3 ${MO_ROOT_PATH}/mo_Caffe*.py --input_model <path_to_model]/VGG16_faster_rcnn_final.Caffe*model --input_proto <path_to_model]/deploy.prototxt --extensions <path_to_object_detection_sample]/fasterrcnn_extensions

 

0 Kudos
1 Solution
Severine_H_Intel
Employee
136 Views

Dear Zhiqi, 

in our previous R1 release of OpenVINO, we had to give the path to the custom layer extension files for faster rcnn: python3 ${MO_ROOT_PATH}/mo_Caffe*.py --input_model <path_to_model]/VGG16_faster_rcnn_final.Caffe*model --input_proto <path_to_model]/deploy.prototxt --extensions <path_to_object_detection_sample]/fasterrcnn_extensions

This is not necessary with our newest release that I invite you to update to. Then, with your command line, it should work.

Best, 

Severine

View solution in original post

3 Replies
Severine_H_Intel
Employee
137 Views

Dear Zhiqi, 

in our previous R1 release of OpenVINO, we had to give the path to the custom layer extension files for faster rcnn: python3 ${MO_ROOT_PATH}/mo_Caffe*.py --input_model <path_to_model]/VGG16_faster_rcnn_final.Caffe*model --input_proto <path_to_model]/deploy.prototxt --extensions <path_to_object_detection_sample]/fasterrcnn_extensions

This is not necessary with our newest release that I invite you to update to. Then, with your command line, it should work.

Best, 

Severine

View solution in original post

Yavorski__Nick
Beginner
136 Views

I have trouble if you find solution please write me

liu__zhiqi
Beginner
136 Views

I upgraded to R3, and I don't have this issue any more. Thank you.

Reply