Performance of face recognition model

[SABRINASST]

Hello community,

I am encountering an issue while attempting to convert my custom ONNX model to an OpenVINO representation.

Description:

I conducted tests on my custom ONNX model for face recognition and obtained the correct results. However, when I attempted to convert this ONNX model into an intermediate OpenVINO representation, I ended up with incorrect results.

I'm uncertain about the exact nature of the problem. Below,
you can find the details of my custom ONNX model.

I kindly request your assistance in resolving this issue. 

THANKS IN ADVANCE

[Aznie_Intel]

Hi Sabrina Sst,

Thanks for reaching out.

Your model input data is dynamic and you have to provide the --input_shapes when converting the model into an Intermediate Representation (IR) file.

Below is the error from Benchmark_App when input shape is not specified. Benchmark_App cannot execute the inference when input is not specify.

Both ONNX and IR file:

Convert your model into IR using the command below:

mo --input_model Downloads\arcface_mobilefacenet\mbf.onnx --input_shape [1,3,112,112]

Below is the result from Benchmark_App after converting with input shape parameter.

1) benchmark_app -m Downloads\arcface_mobilefacenet\mbf.onnx --input_shape [1,3,112,112]

2) benchmark_app -m mbf.xml (*Use IR files generated with input shape)

Hope this helps.

Regards,

Aznie

[Aznie_Intel]

Hi Sabrina SST,

This thread will no longer be monitored since we have provided a solution. If you need any additional information from Intel, please submit a new question. 

Regards,

Aznie