Reproduce “Random Walk on Pixel Manifolds for Anomaly Segmentation of Complex Driving Scenes” 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
Zelong Zeng, Kaname Tomite
RWPM refines pixel features at inference time for anomaly segmentation, letting similar pixels exchange information before classification. It uses a random-walk update without extra training or network changes, though it adds computation. With RbA, it raises average AP across four benchmark settings from 84.73 to 86.00 and lowers FPR95 from 6.33 to 5.46, but does not lead every track metric.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.