MAXIM in TensorFlow

Hi folks,

I hope y’all are doing well. Today, I am delighted to announce the TensorFlow port of MAXIM [1]. MAXIM introduces a single shared vision backbone capable of performing multiple low-level vision tasks such as deraining, dehazing, denoising, etc.

This release has a total of 11 model variants and comes packed with interactive Space demos, XLA support, and Colabs.


Learn more here:


[1] MAXIM: Multi-Axis MLP for Image Processing

Thanks to @lgusm for helpful discussions.


Thanks for the great work Sayak!!

And thanks for also publishing them on TensorFlow Hub with great documentation, sample code and a notebook to help anyone that wants to play with the models!

Very good!


Good work bro…expecting more real life research from you aead.

Well done! And thank you for sharing @Sayak_Paul