Reproduce “Parallax Portrait Matting” 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
Xin Cai, Jiawen Chen, Lars Jebe, Tianfan Xue, Zhoutong Zhang
Parallax Portrait Matting uses two nearby portrait views to estimate a foreground, alpha matte, and foreground colors for one frame. It aligns the background as primary evidence and uses noisier foreground alignment through feature-space cross-attention. On synthetic PPM-100, its SAD is 4.13 versus 5.53 for MatteFormer; real-photo results are qualitative and depend on useful parallax.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.