Which is best yolo v3 source code for openvino compatibility (to run with object detection demo code)

[Nettleton__David]

Dear Sir/Madam,

Can you recommend the best yolo v3 source code to use to train our models so that they will work with openvino (with the runtime c++ code for the yolo v3 demo)?

Below are the details of our attempts.

Thankyou in advance and best regards,

David N.

----------------------------------------------------------

We wanted to train YOLO v3 model on our custom data set and feed it to model optimizer to be able to use openvino object_detection_demo_yolov3_async.py prediction script.

However, there are many implementations of the yolov3 available in github.

I don't know if the https://github.com/mystic123/tensorflow-yolo-v3 (mentioned in your website) is possible to train on my custom data.

Currently I am using yolov3 version of https://github.com/YunYang1994/tensorflow-yolov3 trained on my custom data.

However, when passed to the model optimizer it produced a lot of errors and with the help of openvino form I was able to solve some of those problems. Currently, the model optimizer is saying that some nodes are not converted and that is due to "FusedBatchNormalizationV3" usage, I know that in order to solve this problem I have to build custom layer.. However, even if I solve this the probability of making this implementation of yolov3 work with openvino is quite less and it would be really helpful if you can provide/suggest any source of yolov3 implementation that I can use to train on my custom data and also that is compatible with openvino.

----------------------------------------------------------

[Luis_at_Intel]

Hi Nettleton, David,

I believe the implemented Yolo-V3 source used by OpenVINO can be found in https://github.com/mystic123/tensorflow-yolo-v3. You should be able to train the model using your custom data and convert it to IR using the Model Optimizer. You can find more info on how to convert a TF model to IR here.

Also the model you are already tried seems to be based of off TF 2.0. Another user had a similar issue and they mentioned that they used TF v1.14 for freezing the model, converting the model w/Model Optimizer and Inferencing worked. Not sure if this will resolve your issue but I wanted to mention it. 

If you need any help converting the model to IR format, don't hesitate to reach back. We can attempt to convert it as well if you would like us to try.

Regards,

Luis

[Nettleton__David]

Thanks, Luis. We'll try your suggestions and let you know how we get on !

Best regards,

David N.

[Brewer__Frank]

Hi David,

I was wondering whether you succeded in training a model and converting it to OpenVino IR with one of those two repos your mentioned.

I am having a problem with converting the model, so if you succeded with one of those, I am planning to give it a shot. :) 

Kind regards,
Frank