计算机应用研究2025，Vol.42Issue(7)：1921-1929,9.DOI:10.19734/j.issn.1001-3695.2024.11.0433
多模态行人重识别研究综述
Review of multimodal pedestrian re-identification
摘要
Abstract
In modern security surveillance systems,pedestrian re-identification technology plays a crucial role.Traditional pe-destrian re-identification systems face limitations in accuracy and reliability due to challenges such as changes in lighting,differences in viewing angles,and occlusions in pedestrian images.To address these challenges,researchers have introduced multi-modal learning methods into the RE-ID field,hoping to effectively integrate various data modalities,such as depth ima-ges,infrared images,and textual information,to improve the performance of RE-ID.This paper reviewed the application and re-search progress of multi-modal RE-ID technology in modern security surveillance systems.It firstly introduced the basic con-cepts of multi-modal technology and the task of multi-modal RE-ID,and then outlined the key datasets and evaluation protocols in this field.The core part of the article discussed in detail the fusion strategies in multi-modal RE-ID,including feature-level fusion and model-level fusion methods.Finally,the article discussed the application scenarios and future research directions of multi-modal RE-ID.关键词
RE-ID/多模态融合/数据融合/特征融合/模型融合Key words
RE-ID/multimodal fusion/data fusion/feature fusion/model fusion分类
信息技术与安全科学引用本文复制引用
石瑞鑫,智敏,殷雁君..多模态行人重识别研究综述[J].计算机应用研究,2025,42(7):1921-1929,9.基金项目
内蒙古自然科学基金资助项目(2023MS06009) （2023MS06009）
内蒙古高等学校科学研究重点资助项目(NJZZ21004) （NJZZ21004）
