Hi!

I recently re-implemented a neural network from this paper: https://doi.org/10.2352/issn.2694-118X.2021.LIM-54 in Python (specifically PyTorch) for my Master's thesis. Unfortunately, I could not use the colour difference functions (e.g. CIEDE2000) implemented in this library, as they operate on Numpy arrays rather than PyTorch tensors.

I had an idea to make an ML library for colour science functions that implement basic conversions and colour difference functions for the most popular frameworks (TensorFlow and PyTorch). At the moment there is no "open-source" library to my knowledge for this, although ML/DL is a growing field. There are individual packages, such as: https://github.com/jorge-pessoa/pytorch-colors and https://github.com/ZhengyuZhao/PerC-Adversarial/tree/master, but I figured it would be worth the effort.

Would this be something you support, maybe as another library in the colour-science family?