Reproduce “LensErase: A Foundation Model for Erase Anything” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Yunuo Chen, Xiuyu Wu, Qinhong Yang, Jinjing Zhao, Yanchen Dong, Sirui Zhang, Dong Chen, Fangyun Wei and 6 more
LensErase removes varied objects, text, watermarks, and associated effects from images using an image and mask, without a text prompt. It keeps the intact source visible and encodes masked and unmasked regions with Mask-Aware RoPE. On a 500-case reference-free benchmark, it scores 95.22 Overall and 82.80% HiFi Removal, versus OmniPaint’s 73.59 and 42.80%; these judgments do not establish 4K performance.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.