Nighttime Person Re-Identification via Collaborative Enhancement Network with Multi-domain Learning

2023-12-25

arXiv.org

AI Summary

To address severe image degradation and the entanglement of illumination enhancement with identity discrimination in nighttime person re-identification (ReID), this paper proposes the Collaborative Enhancement Network (CENet). Our method decouples illumination restoration from ReID via a parallel multi-level feature interaction architecture built upon a shared Transformer encoder. To mitigate low-light feature degradation, we introduce a cross-task feature distillation mechanism. Furthermore, an alternating multi-domain learning algorithm is designed to effectively bridge the domain gap between limited real-world nighttime data and large-scale synthetic data. Extensive experiments demonstrate that CENet achieves state-of-the-art performance on two real-world nighttime benchmarks—Night600 and RGBNT201_rgb—improving mAP by over 8.2% compared to prior methods. The source code and synthetic dataset are publicly released.

Abstract

Prevalent nighttime person re-identification (ReID) methods typically combine image relighting and ReID networks in a sequential manner. However, their performance (recognition accuracy) is limited by the quality of relighting images and insufficient collaboration between image relighting and ReID tasks. To handle these problems, we propose a novel Collaborative Enhancement Network called CENet, which performs the multilevel feature interactions in a parallel framework, for nighttime person ReID. In particular, the designed parallel structure of CENet can not only avoid the impact of the quality of relighting images on ReID performance, but also allow us to mine the collaborative relations between image relighting and person ReID tasks. To this end, we integrate the multilevel feature interactions in CENet, where we first share the Transformer encoder to build the low-level feature interaction, and then perform the feature distillation that transfers the high-level features from image relighting to ReID, thereby alleviating the severe image degradation issue caused by the nighttime scenario while avoiding the impact of relighting images. In addition, the sizes of existing real-world nighttime person ReID datasets are limited, and large-scale synthetic ones exhibit substantial domain gaps with real-world data. To leverage both small-scale real-world and large-scale synthetic training data, we develop a multi-domain learning algorithm, which alternately utilizes both kinds of data to reduce the inter-domain difference in training procedure. Extensive experiments on two real nighttime datasets, extit{Night600} and extit{RGBNT201$_{rgb}$}, and a synthetic nighttime ReID dataset are conducted to validate the effectiveness of CENet. We release the code and synthetic dataset at: hyperlink{https://github.com/Alexadlu/CENet}{color{red} https://github.com/Alexadlu/CENet}.

Problem

• Low-light Recognition
• Image Brightness Improvement
• Visual Recognition Precision

Innovation

• CENet
• Multi-level Feature Exchange
• Nighttime Image Recognition