Reproduce “Confidence-Guided Pseudo-Label Refinement for Unsupervised Cross-Domain Person Re-Identification” 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
Sundas Iqbal, Danish Ali, Qing Tian, Yuhui Zheng, Jianping Gou, Weihua Ou
CGPLR adapts person re-identification models to unlabeled camera datasets by refining cluster-generated identity labels. It combines centroid fit, neighborhood agreement, separation from competing clusters, and temporal stability to preserve reliable labels, reassign supported alternatives, or defer uncertain samples. On Market-1501→MSMT17, it reports 76.4% Rank-1 and 47.6% mAP, versus ADDNet’s 75.2% and 46.7%; results use published protocols, not a controlled head-to-head.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.