LAF: Enhancing person re-identification via Latent-Assisted Feature Fusion

Minglang Li, Zhiyong Tao, Sen Lin, Kaihao Feng

Abstract: Highlights•Designed LA-FUSION, an auxiliary network for latent information extraction via guided feature fusion.•Proposed improved fusion strategy integrating shallow-deep features for person re-identification.•Dedicated latent info extraction scheme: primary feature-guided erasing followed by enhancement.

External IDs:doi:10.1016/j.aej.2025.04.075