陈元妹,王凤随,王路遥.细化特征引导对抗性解纠缠学习的无监督行人重识别[J].电子测量与仪器学报,2024,38(5):130-138 |
细化特征引导对抗性解纠缠学习的无监督行人重识别 |
Unsupervised person re-identification of adversarialdisentangling learning guided by refined features |
|
DOI: |
中文关键词: 行人重识别 无监督 特征细化 相机差异 对抗性解纠缠 |
英文关键词:person re-identification unsupervised feature refinement camera differences adversarial disentangling |
基金项目:安徽省自然科学基金(2108085MF197)、安徽高校省级自然科学研究重点项目(KJ2019A0162)、安徽工程大学国家自然科学基金预研项目(Xjky2022040)资助 |
|
Author | Institution |
Chen Yuanmei | 1.School of Electrical Engineering, Anhui Polytechnic University, Wuhu 241000, China; 2.Key Laboratory of
Advanced Perception and Intelligent Control of Highend Equipment, Ministry of Education, Wuhu 241000, China |
Wang Fengsui | 1.School of Electrical Engineering, Anhui Polytechnic University, Wuhu 241000, China; 2.Key Laboratory of
Advanced Perception and Intelligent Control of Highend Equipment, Ministry of Education, Wuhu 241000, China |
Wang Luyao | 1.School of Electrical Engineering, Anhui Polytechnic University, Wuhu 241000, China; 2.Key Laboratory of
Advanced Perception and Intelligent Control of Highend Equipment, Ministry of Education, Wuhu 241000, China |
|
摘要点击次数: 1 |
全文下载次数: 274 |
中文摘要: |
无监督行人重识别旨在无监督设置下从非重叠的相机中识别出同一行人。针对现有的无监督行人重识别网络不能充分提取行人特征以及相机之间的差异导致行人检索错误的问题,提出了一种细化特征引导对抗性解纠缠学习的无监督行人重识别方法,设计特征细化信息融合模块嵌入ResNet50网络的不同层,用以增强网络提取关键信息的能力。设计特征解耦学习方法最小化行人特征和相机特征之间的互信息,减少相机差异对网络的负面影响,同时设计对抗性解纠缠损失函数进行无监督联合学习。在两个公共数据集Market-1501和DukeMTMC-reID上对所提方法进行评估,平均精度均值分别提升了4.6%、3.1%,相较于基线算法具备较强的鲁棒性,满足在无监督背景下对行人的识别需求。 |
英文摘要: |
Unsupervised person re-identification aims to identify the same person from non-overlapping cameras under unsupervised settings. Aiming at the problem that the existing unsupervised person re-identification network cannot fully extract pedestrian features and the difference between cameras leads to pedestrian retrieval errors, we propose an unsupervised person re-identification of adversarial disentangling learning guided by refined features. A feature refinement information fusion module is designed and embedded into different layers of ResNet50 network to enhance the ability of the network to extract key information. A disentangled feature learning method is designed to minimize the mutual information between pedestrian features and camera features, and reduce the negative impact of camera differences on the network. At the same time, the adversarial disentangling loss function is designed for unsupervised joint learning. Using the Market-1501 and DukeMTMC-reID public datasets, we tested the proposed method. The mean average precision increased by 4.6% and 3.1% respectively. Compared with the baseline algorithm, it has strong robustness and meets the needs of pedestrian recognition in unsupervised background. |
查看全文 查看/发表评论 下载PDF阅读器 |
|
|
|