Skip to content

Latest commit

 

History

History
28 lines (19 loc) · 1.28 KB

README.md

File metadata and controls

28 lines (19 loc) · 1.28 KB

RapidEsrGan

An implementation of ESRGAN using WebNN See https://sushanthr.github.io/RapidEsrGan/

Fully offline, client side super resolution in your browser.

Model details

Original model - https://github.com/xinntao/Real-ESRGAN

Note on Browser Requirement (Chrome/Edge Canary on Windows)

WebNN is a new standard, in development.
Check https://webmachinelearning.github.io/webnn-status/ for updates on your fav platform/browser.

For now the latest canary of any chromium based browser:
Chrome/Edge Canary on Windows has enough of the spec implemented for this implementation to work.

You would also need the experimental WebNN API enabled, paste the following URL in the address bar to get to that settings page.

chrome://flags/#web-machine-learning-neural-network

Please pay attention to the warning WebNN is an experimental API, proceed at your own risk.
WARNING: EXPERIMENTAL FEATURES AHEAD! By enabling these features, you could lose browser data or compromise your security or privacy. Enabled features apply to all users of this browser. If you are an enterprise admin you should not be using these flags in production.

Chrome Canary https://www.google.com/chrome/canary/
Edge Canary https://www.microsoft.com/en-us/edge/download/insider?form=MA13FJ