Read the original at HF Daily Papers
Researchers introduce LensErase, a general-purpose foundation model for removing objects, text, and watermarks from images. It is trained on a dataset containing 101.07 million paired examples.
Carried by: HF Daily Papers. First seen: .