Intel® Distribution of OpenVINO™ Toolkit
Community support and discussions about the Intel® Distribution of OpenVINO™ toolkit, OpenCV, and all things computer vision-related on Intel® platforms.

Converting Yolov5 failed with weird errors

HenryFuhengWu
Beginner
367 Views

I was trying to convert a YOLOv5 model from onnx to OV IR but got the error as below:

HenryFuhengWu_0-1631721353949.png

This original model is from official Yolov5 repo. Anyone can help? Thanks.

 

0 Kudos
4 Replies
Peh_Intel
Moderator
344 Views

Hi Henry,

 

Thanks for reaching out to us.

 

For your information, YOLOv5 is currently not an officially supported topology by the OpenVINO toolkit.

 

However, I would like to share with you about this tutorial.

https://github.com/violet17/yolov5_demo.

 

Identify the output nodes of your model with Netron and specify them in the Model Optimizer.

 

For example:

 

python mo_onnx.py --input_model yolov5m.onnx -s 255 --reverse_input_channels --output Conv_322,Conv_375,Conv_428

 

 

Regards,

Peh

 

HenryFuhengWu
Beginner
335 Views

Thanks for the reply, Peh. I don't understand why the YOLOV5 is not supported. I feel it may be impossible to enumerate all the models in the world and add support to them one by one... Say if I have my own model called XYZ, how do I know if OpenVino supports it or not? Can you share me the root cause of not supporting it? Is it because of some technical challenges or some IR design limitation, or even hardware issue or whatever? Sorry for trying to dig deep. Thanks a lot.

Peh_Intel
Moderator
318 Views

Hi Henry,

 

Yolo v5 was stated not officially supported by the OpenVINO toolkit because it yet to validated by Developer teams and hence include the model in the Public Pre-trained Models.


To ensure a model can be supported by OpenVINO toolkit, you can check whether the model is one of the supported topologies and its framework layer is supported in Supported Framework Layers. If one of the layers in the model is not in the supported framework layers list, the Model Optimizer might unable to recognize that layer and hence fails when converting into Intermediate Representation (IR).



Regards,

Peh


Peh_Intel
Moderator
272 Views

Hi Henry,


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



Regards,

Peh


Reply