LaMa-style image inpainting, fully on-device, for iOS. The same "magic erase" UX you've seen in Photos, running on your iPhone with zero server.
By Daisuke Majima. The LaMa weights are ported in CoreML-Models (1,749★). This repo wraps them in a usable iOS sample.
📱 Powers Mask2Face on the App Store.
💼 Open to Staff / Senior iOS / Mobile ML roles — john-rocky.github.io
1, Clone this repository(xcode project).
2, Download LaMa.mlpackage from CoreML-Models.
3, Drag and drop LaMa.mlpackage to the xcode project.
4, Run the xcode project.
Model Repository: advimman/lama
Concersion Script: mallman/CoreMLaMa
Inspired by: Sanster/lama-cleaner

