Reproduce “Leveraging 2D Foundation Priors for Generalized Category Discovery under Test-Time Adaptation in 3D LiDAR Segmentation” 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
Hanqi Chen, Chaoqi Chen, Tian Ji, Hui Huang
The authors present a LiDAR method for adapting to unlabeled target streams containing known and novel categories. It filters known-class pseudo-labels using local geometry and class prototypes, balances assignments to novel categories, and gates image features into LiDAR representations. It ranks first in All mIoU on 9/10 synthetic-to-real splits, but assumes the novel-category count is known.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.