GLCN: Graph-Aware Locality-Enhanced Cross-Modality Re-ID Network

TL;DR

This paper introduces GLCN, a new network for cross-modality person re-identification that improves accuracy by enhancing local features and aligning structures across different modalities.

Contribution

The novel GLCN framework introduces modules for locality preservation, cross-branch consistency, and geometry alignment, along with a new loss function for compact identity structures.

Findings

GLCN outperforms existing methods on benchmarks like SYSU-MM01 and RegDB.

The proposed modules and loss function effectively reduce intra-class variance and improve inter-class separation.

The framework enhances cross-modality alignment and robustness to illumination and occlusion issues.

Abstract

Cross-modality person re-identification faces challenges such as illumination discrepancies, local occlusions, and inconsistent modality structures, leading to misalignment and sensitivity issues. We propose GLCN, a framework that addresses these problems by enhancing representation learning through locality enhancement, cross-modality structural alignment, and intra-modality compactness. Key components include the Locality-Preserved Cross-branch Fusion (LPCF) module, which combines Local–Positional–Channel Gating (LPCG) for local region and positional sensitivity; Cross-branch Context Interpolated Attention (CCIA) for stable cross-branch consistency; and Graph-Enhanced Center Geometry Alignment (GE-CGA), which aligns class-center similarity structures across modalities to preserve category-level relationships. We also introduce Intra-Modal Prototype Discrepancy Mining Loss (IPDM-Loss) to reduce intra-class variance and improve inter-class separation, thereby creating more compact identity structures in both RGB and IR spaces. Extensive experiments on SYSU-MM01, RegDB, and other benchmarks demonstrate the effectiveness of our approach.

Figures