Reproduce “View-decoupled Transformer for Person Re-identification under Aerial-ground Camera Network” 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

Quan Zhang, Lei Wang, Vishal M. Patel, Xiaohua Xie, Jianhuang Lai

The authors introduce VDT, a person re-identification method for matching identities across aerial and ground cameras. At each transformer layer, it subtracts a learned view representation from a global representation and trains them to be less alike. On real AG-ReID ground-to-aerial retrieval, VDT reaches 86.59% Rank-1 and 78.57% mAP, exceeding ViT; evaluations remain limited to two datasets, including synthetic CARGO.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.