[Detection] About changing train function

Hi

I have a question about retraining models with another train function. Can I change internal structure like backbone network, loss function, optimizer and so on in baseline model which is provided by github repo. Thanks for your attention

dowan

Hi Dowan,

Thanks for your question. I am sorry that this was not clear in the tutorial: but of course, indeed you should!

The baseline repo is only to provide an example structure, but you should modify it according to your own method. So in your test and train function, you should implement functions to train and test your own detection model.

Best, Ecem