Computer Vision Deep Learning loss functions

A question to Computer Vision Deep Learning people, what are the CV specific loss functions that are worth investigating. Like ‘standard’ ones, but also new research from papers, etc. Is there any resource dedicated to explaining how they work and when they are appropriate (structure of data, use case, etc.) and why? Thanks!!!

I recently stumbled over this repo with a focus on segmentation and the overview picture with the different losses is great:


I’m doing segmentation and there’s a whole host of loss functions but not all are relevant for other tasks.
Thus you might want to divide this thread into loss functions for different tasks.

I’m exploring structured similarity loss today for example for segmentation.

@MicPie - that’s a great repo

1 Like

This is great! Thank you :heart_eyes:

That’s a good idea. Thanks!

1 Like