Covnert PyTorch-YOLOv3 to Intel IR

[Bigishios__Gjegj]

Hi all.

Is it possible to convert PyTorch-YOLOv3 to Intel IR?

I found instruction just for a TensorFlow implementation

[Iffa_Intel]

Greetings,

You will found the documentation regarding converting YOLO model to Intel IR here:

http://docs.openvinotoolkit.org/latest/_docs_MO_DG_prepare_model_convert_model_tf_specific_Convert_YOLO_From_Tensorflow.html

This tutorial explains how to convert real-time object detection YOLOv1*, YOLOv2*, and YOLOv3* public models to the Intermediate Representation (IR).

Sincerely,

Iffa

[Iffa_Intel]

Greetings,

I may have missed you were specifically referring to a PyTorch implementation of YoloV3 and not a TensorFlow implementation of YoloV3.

PyTorch is not officially supported by OpenVINO, but you can try converting  it into ONNX format and then to IR with Model Optimizer.

The guide here covers converting PyTorch models to IR via ONNX conversion, however as the guide shows, YoloV3 is not under the supported models (OpenVINO™ toolkit officially supports public Pytorch* models (from torchvision 0.2.1 and pretrainedmodels 0.7.4 packages) via ONNX conversion).

You can refer here: 

http://docs.openvinotoolkit.org/2018_R5/_docs_MO_DG_prepare_model_convert_model_Convert_Model_From_ONNX.html

Hope this helps!

Sincerely,

Iffa

[Iffa_Intel]

Hi,

I'm closing the case since the are no reply within 5 working days.

If there are any inquiries please help to create a new thread.

Thanks & have a great day ahead.

Sincerely,

Iffa