View-Aware Semantic Alignment for Aerial-Ground Person Re-Identification

2026-05-18

AI Summary

This work addresses the challenge of person re-identification across drastically different viewpoints between drones and ground cameras, a setting where existing methods often neglect view-specific cues. To overcome this limitation, the authors propose the View-aware Semantic Alignment (ViSA) framework, which departs from conventional view-invariant paradigms by introducing a view-aware mechanism that effectively leverages view-specific features while preserving identity-discriminative information. ViSA employs an Expert-driven Token Generation Module (ETGM) to construct adaptive semantic queries and integrates a Dual-branch Local Fusion Module (DLFM) with graph-based reasoning to achieve semantic alignment of local regions. The method achieves state-of-the-art performance on three benchmarks—AG-ReID.v2, CARGO, and LAGPeR—with a notable 10.06% mAP improvement under the cross-view protocol on CARGO.

Abstract

Aerial-Ground Person Re-Identification (AGPReID) remains highly challenging due to drastic viewpoint variations between drones and fixed cameras. Existing methods typically follow a view-invariant paradigm, aligning shared features across views to achieve robustness. However, view-invariant inherently enforces part-level alignment, which ignores view-specific cues and discriminative identity information. To this end, this work proposes ViSA (View-aware Semantic Alignment), a view-aware framework that achieves cross-view semantic consistency containing an Expert-driven Token Generation Module (ETGM) and a Dual-branch Local Fusion Module (DLFM). Technically, the former constructs a set of view-aware experts to generate adaptive semantic queries that perceive viewpoint-specific patterns, while the latter leverages graph reasoning to extract and align local regions responsive to different experts. Extensive experiments on three AGPReID benchmarks including AG-ReID.v2, CARGO and LAGPeR demonstrate that ViSA consistently achieves superior performance, with a notable 10.06\% mAP improvement on the challenging CARGO cross-view protocol. The code is available at \href{https://github.com/Cat-Zero/ViSA}{https://github.com/Cat-Zero/ViSA}.

Problem

• Aerial-Ground Person Re-Identification
• viewpoint variation
• view-invariant
• semantic alignment
• cross-view

Innovation

• View-aware Semantic Alignment
• Expert-driven Token Generation
• Dual-branch Local Fusion
• Graph Reasoning
• Aerial-Ground Person Re-Identification