Balancing within-person and between-person variation improves clothes-changing re-identification

MINER adaptively balances intra- and inter-person variance to achieve state-of-the-art results on multiple clothes-changing benchmarks

Big Tech
Oron Nir · Eliyahu Strugo · Ariel Shamir

Reichman University · Microsoft

Research Digest··2 min read
Researchers from Reichman University and Microsoft introduce MINER, a model-agnostic training framework that explicitly balances intra- and inter-person variance for Clothes-Changing Person Re-Identification (CC-ReID).

The authors formulate CC-ReID as a trade-off between inter-person discrimination and intra-person clothing invariance.

§