FUSE: Frequency-domain Unification and Spectral Energy Alignment for Multi-modal Object Re-Identification

2026-06-18

AI Summary

Existing multimodal person re-identification methods overly rely on low-frequency information while neglecting mid- and high-frequency structural details, resulting in incomplete representations and unstable cross-modal alignment. This work proposes the first unified frequency-domain framework that formulates the task as a two-stage process of spectral disentanglement and energy alignment. By adaptively partitioning the frequency subspace, the approach explicitly models and complementarily aligns low-, mid-, and high-frequency components, further enhanced by learnable frequency-domain modulation for improved robustness. The designed Spectral Decomposition Module (SDM) and Cross-modal Alignment Module (CAM), combined with frequency-domain consistency regularization, achieve significant performance gains—improving mAP by 9.1% and Rank-1 accuracy by 9.5% on RGBNT201, RGBNT100, and MSVR310 benchmarks.

Abstract

Despite significant progress in multi-modal Re-Identification (ReID), existing methods tend to emphasize low-frequency cues. Consequently, they focus on attributes such as color, illumination, and coarse appearance, while overlooking mid and high-frequency structures that encode geometric, textural, and identity-discriminative details. This imbalance leads to incomplete spectral representations and unstable cross-modal alignment. To overcome these limitations, we introduce FUSE, a frequency-domain framework that reformulates multi-modal ReID as a two-stage process of spectral disentanglement and energy alignment. The proposed Spectral Decomposition Module (SDM) adaptively partitions features into low, mid, and high-frequency subspaces, enabling hierarchical spectral modeling. The Cross-Modal Alignment Module (CAM) further enforces energy alignment and subspace complementarity across modalities via frequency-consistency regularization. In addition, FUSE incorporates learnable frequency modulation to enhance robustness under varying illumination and heterogeneous sensor conditions. Extensive experiments on RGBNT201, RGBNT100, and MSVR310 show that FUSE achieves 9.1\% mAP and 9.5\% Rank-1 improvements, establishing an interpretable frequency-domain paradigm for multi-modal representation learning.

Problem

• multi-modal Re-Identification
• frequency-domain representation
• spectral imbalance
• cross-modal alignment
• high-frequency details

Innovation

• frequency-domain representation
• spectral decomposition
• cross-modal alignment
• multi-modal ReID
• frequency modulation