Exploring Modality-shared Appearance Features and Modality-invariant Relation Features for Cross-modality Person Re-Identification

TL;DR

This paper introduces a novel approach for cross-modality person re-identification that combines modality-shared appearance features with modality-invariant relation features, utilizing a multi-level two-stream network and a new quadruplet loss to improve accuracy.

Contribution

The paper proposes a multi-level two-stream network that captures both appearance and relation features, along with a novel quadruplet loss, to enhance cross-modality person re-identification performance.

Findings

Outperforms state-of-the-art methods on benchmark datasets.

Effectively reduces cross-modality and intra-modality variations.

Improves identification accuracy in RGB and infrared images.

Abstract

Most existing cross-modality person re-identification works rely on discriminative modality-shared features for reducing cross-modality variations and intra-modality variations. Despite some initial success, such modality-shared appearance features cannot capture enough modality-invariant discriminative information due to a massive discrepancy between RGB and infrared images. To address this issue, on the top of appearance features, we further capture the modality-invariant relations among different person parts (referred to as modality-invariant relation features), which are the complement to those modality-shared appearance features and help to identify persons with similar appearances but different body shapes. To this end, a Multi-level Two-streamed Modality-shared Feature Extraction (MTMFE) sub-network is designed, where the modality-shared appearance features and modality-invariant relation features are first extracted in a shared 2D feature space and a shared 3D feature space, respectively. The two features are then fused into the final modality-shared features such that both cross-modality variations and intra-modality variations can be reduced. Besides, a novel cross-modality quadruplet loss is proposed to further reduce the cross-modality variations. Experimental results on several benchmark datasets demonstrate that our proposed method exceeds state-of-the-art algorithms by a noticeable margin.