How to train ReID model and convert to .onnx ?

See original GitHub issue

Hi @GeekAlexis , For the customs classes (person and car), training a Yolo model is very simple, but can you explain how to train ReID model? Can you please provide the link for the exact project you used to train the current DeepSort ReID model? Can we use the same project for training model for person car? I swapped the feature extractor in Deep SORT for a better ReID model, OSNet.. Do you mean you are using OSNet in place of DeepSort? Just a little bit confusion the OSNet is another tracker or some kind of implementation of DeepSort?

My goal is to track person and car. I know how to train Yolo model but the ReID step is not clear. Please explain when you have some spare time. Thanks.

Issue Analytics

State:
Created 3 years ago
Comments:7 (7 by maintainers)

Top GitHub Comments

1reaction

GeekAlexiscommented, Nov 6, 2020

You can do it within pytorch. Use torch.onnx.export

On Thu, Nov 5, 2020 at 10:19 PM Asad notifications@github.com wrote:

@GeekAlexis https://github.com/GeekAlexis I want to want to ask how did you convert OSNet ReID model to .onnx? I am planning to convert ResNet ReID model to .onnx and then try it for one class, if it can run then I will train ResNet for two classes.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/GeekAlexis/FastMOT/issues/22#issuecomment-722883639, or unsubscribe https://github.com/notifications/unsubscribe-auth/AGNJO5UQIGH5KNEGLEX7YMTSOOIP5ANCNFSM4THHNEPA .

0reactions

MuhammadAsadJavedcommented, Nov 6, 2020

@GeekAlexis I want to ask how did you convert OSNet ReID model to .onnx? I am planning to convert ResNet ReID model to .onnx and then try it for one class, if it can run then I will train ResNet for two classes.