Yolo TAO non maximum suppresion (NMS)

Hi, From the original Yolo via darknet I used to be able to set NMS while using already trained model.
Now I work with Yolov4 TAO, and I see no way to do it. I only see this option inside the spesc txt file being used during training process.
Thus my question is: Is it possible in YOLO Tao to set NMS while using already trained model? If yes, how?

It is not just used in training process. NMS config applies to the NMS layer of the model in training, validation, evaluation, inference, and export.

Hi thank you very much for the answer.
Yes, I know that NMS is needed during the training, yet, my question was about setting nms while running the already trained network.
Actually I found a way to set nms during rutine running. I ve noticed that to run an inference it needs a path to specs txt file which was used during the training. I have changed the nms settings there (after the network was trained) and it worked.
Still remains a question is it the way to do it?
Such a way has one great disadvantage, it is unavailable in the stage i use the exported etlt model

There is no update from you for a period, assuming this is not an issue anymore.
Hence we are closing this topic. If need further support, please open a new one.

Yes, you can change the config when run evaluation, inference and export.

For exporting to .etlt model, you can still use/modify the NMS config in the spec file.

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.