Reproduce “DC-Former: Diverse and Compact Transformer for 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
Wen Li, Cheng Zou, Meng Wang, Furong Xu, Jianan Zhao, Ruobing Zheng, Yuan Cheng, Wei Chu
DC-Former addresses cross-camera person re-identification by learning multiple class-token descriptions of each image, concatenated for retrieval. A self-diverse constraint (SDC) encourages tokens to separate. On MSMT17, it reports 70.7 mAP and 86.9 Rank-1, versus TransReID’s 69.4 and 86.2; this matched-resolution comparison does not establish lower computational cost or prove the authors’ identity-density hypothesis.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.