search

keras-team / keras-cv

Industry-strength Computer Vision workflows with Keras
Other
973 stars 317 forks source link

Super-resolution using diffusion model #1910

Open shshrzad opened 1 year ago

shshrzad commented 1 year ago

Hello;

Thank you for your fantastic works in keras-cv. I want to train a super-resolution model based on diffusion to enlarge and enhance the images in my dataset. Can I use stable diffusion that you provide in this repository for this purpose? How should I train this model? should I prepare joint HR-LR images for this end? I will appreciate you if guide me.

jbischof commented 12 months ago

@shshrzad thanks for reaching out! Unfortunately our Stable Diffusion model is not fully modularized to allow you to pick off a single component like this. If you're interested in contributing such a model that would awesome!