Filters








23,177 Hits in 4.7 sec

Dual Cluster Contrastive learning for Person Re-Identification [article]

Hantao Yao, Changsheng Xu
2021 arXiv   pre-print
Recently, cluster contrastive learning has been proven effective for person ReID by computing the contrastive loss between the individual feature and the cluster memory.  ...  Therefore, we formulate the individual-based updating and centroid-based updating mechanisms in a unified cluster contrastive framework, named Dual Cluster Contrastive learning (DCC), which maintains two  ...  Dual Cluster Contrastive learning for Person Re-Identification Hantao  ... 
arXiv:2112.04662v2 fatcat:jdgaunbkhjdt3hflltjftd5n2a

Dual-Stream Reciprocal Disentanglement Learning for Domain Adaptation Person Re-Identification [article]

Huafeng Li, Kaixiong Xu, Jinxing Li, Guangming Lu, Yong Xu, Zhengtao Yu, David Zhang
2021 arXiv   pre-print
Since human-labeled samples are free for the target set, unsupervised person re-identification (Re-ID) has attracted much attention in recent years, by additionally exploiting the source set.  ...  To tackle this problem, in this paper we propose a novel method named Dual-stream Reciprocal Disentanglement Learning (DRDL), which is quite efficient in learning domain-invariant features.  ...  Inspired by this, in this paper a novel dual-stream reciprocal disentanglement learning (DRDL) is proposed to simply but efficiently learn domain-invariant features for unsupervised person Re-ID.  ... 
arXiv:2106.13929v2 fatcat:r54cgt6c7nbediu4wumb4vaowi

Successive Embedding and Classification Loss for Aerial Image Classification [article]

Jiayun Wang, Patrick Virtue, Stella X. Yu
2019 arXiv   pre-print
Visualizations of the network's embedded representations reveal that the embedding loss encourages greater separation between target class clusters for both training and testing partitions of two aerial  ...  We demonstrate that networks trained with this dual embedding and classification loss outperform networks with classification loss only.  ...  Zhirong Wu, Baladitya Yellapragada and Michele Winter for valuable discussions and feedback.  ... 
arXiv:1712.01511v3 fatcat:cvzani6zbzd63ez5sie4owopg4

Graph Sampling Based Deep Metric Learning for Generalizable Person Re-Identification [article]

Shengcai Liao, Ling Shao
2021 arXiv   pre-print
Recent studies show that, both explicit deep feature matching as well as large-scale and diverse training data can significantly improve the generalization of person re-identification.  ...  Together with an adapted competitive baseline, we improve the previous state of the art in generalizable person re-identification significantly, by up to 24% in Rank-1 and 13.8% in mAP.  ...  Bai et al. proposed a dual-meta generalization network and a large-scale dataset called Person30K for person re-identification [2] .  ... 
arXiv:2104.01546v3 fatcat:jb7da5cs35e3reugbcrartrbbu

Imitating Targets from all sides: An Unsupervised Transfer Learning method for Person Re-identification [article]

Jiajie Tian, Zhu Teng, Rui Li, Yan Li, Baopeng Zhang, Jianping Fan
2021 arXiv   pre-print
intra-dataset difference via a proposed ImitateModel simultaneously; 2) regarding the unsupervised person Re-ID problem as a semi-supervised learning problem formulated by a dual classification loss to  ...  Person re-identification (Re-ID) models usually show a limited performance when they are trained on one dataset and tested on another dataset due to the inter-dataset bias (e.g. completely different identities  ...  Supervised Learning for Person Re-ID To obtain a good performance in person re-identification task, the prime goal is to learn discriminative representations to distinguish person identities.  ... 
arXiv:1904.05020v2 fatcat:d7tmspdg4fbjngwdrhsaq5gy7a

Refining Pseudo Labels for Unsupervised Domain Adaptive Person Re-Identification

Limin Xia, Zhimin Yu, Wentao Ma, Jiahui Zhu
2021 IEEE Access  
In this paper, we propose a novel framework to refine pseudo labels for UDA person re-ID.  ...  Unsupervised domain adaptive (UDA) person re-identification (re-ID) aims to generalize the model trained on a labeled source domain to an unlabeled target domain.  ...  INTRODUCTION Person re-identification (re-ID) is a task to match the same person across non-overlapping cameras.  ... 
doi:10.1109/access.2021.3108879 fatcat:55yyqnzdwfe5nc5yag64ctzyii

Attentive WaveBlock: Complementarity-enhanced Mutual Networks for Unsupervised Domain Adaptation in Person Re-identification and Beyond [article]

Wenhao Wang, Fang Zhao, Shengcai Liao, Ling Shao
2021 arXiv   pre-print
Unsupervised domain adaptation (UDA) for person re-identification is challenging because of the huge gap between the source and target domain.  ...  Experiments demonstrate that the proposed method achieves state-of-the-art performance with significant improvements on multiple UDA person re-identification tasks.  ...  Acknowledgment We thank Informatization Office of Beihang University for the supply of High Performance Computing Platform.  ... 
arXiv:2006.06525v3 fatcat:q42sixylbredjepd3trokxofcu

G2DA: Geometry-Guided Dual-Alignment Learning for RGB-Infrared Person Re-Identification [article]

Lin Wan, Zongyuan Sun, Qianyan Jing, Yehansen Chen, Lijing Lu, Zhihang Li
2021 arXiv   pre-print
RGB-Infrared (IR) person re-identification aims to retrieve person-of-interest from heterogeneous cameras, easily suffering from large image modality discrepancy caused by different sensing wavelength  ...  In this paper, we propose a graph-enabled distribution matching solution, dubbed Geometry-Guided Dual-Alignment (G2DA) learning, for RGB-IR ReID.  ...  Index Terms-Person re-identification, cross-modality matching, optimal transport, feature alignment, channel exchange. I.  ... 
arXiv:2106.07853v2 fatcat:3nspm3ymgng67esqksxrfqugme

Uncertainty-aware Clustering for Unsupervised Domain Adaptive Object Re-identification [article]

Pengfei Wang, Changxing Ding, Wentao Tan, Mingming Gong, Kui Jia, Dacheng Tao
2021 arXiv   pre-print
State-of-the-art object Re-ID approaches adopt clustering algorithms to generate pseudo-labels for the unlabeled target domain.  ...  Unsupervised Domain Adaptive (UDA) object re-identification (Re-ID) aims at adapting a model trained on a labeled source domain to an unlabeled target domain.  ...  Contrastive Learning As a promising paradigm of unsupervised learning, contrastive learning has lately achieved state-ofthe-art performance in unsupervised visual representation learning.  ... 
arXiv:2108.09682v1 fatcat:7kxdkzzf3fevfdcxjmt6eox3nu

Weighted Cluster-Range Loss and Criticality-Enhancement Loss for Speaker Recognition

Jianye Mo, Li Xu
2020 Applied Sciences  
Additionally, a Weighted Cluster-Range Loss (WCRL) is proposed to enhance the identification performance of Cluster-Range Loss (CRL) on indecisive samples.  ...  While traditional i-vector based methods are popular in the field of speaker recognition, deep learning has recently found more and more applications to the end-to-end models due to its attractive performance  ...  Bian for his contribution to the basic work of the project and the support from Robotics Institute of Zhejiang University. Conflicts of Interest: The authors declare no conflict of interest.  ... 
doi:10.3390/app10249004 fatcat:x3ey7sw5snhjvhejjlwsmradyi

Improving Slice-Based Model for Person Re-ID with Multi-Level Representation and Triplet-Center Loss

Yusheng ZHANG, Zhiheng ZHOU, Bo LI, Yu HUANG, Junchu HUANG, Zengqun CHEN
2019 IEICE transactions on information and systems  
Person Re-Identification has received extensive study in the past few years and achieves impressive progress.  ...  Our model consists of a dualbranch network architecture, one branch for global feature extraction and the other branch for local ones.  ...  Dual Branch Dual-branch setting is very common for person Re-ID tasks.  ... 
doi:10.1587/transinf.2019edp7067 fatcat:nwedriuksbhb7lwbv3fufmcs4q

DomainMix: Learning Generalizable Person Re-Identification Without Human Annotations [article]

Wenhao Wang, Shengcai Liao, Fang Zhao, Cuicui Kang, Ling Shao
2021 arXiv   pre-print
learning between domain-invariant feature learning and domain discrimination, and meanwhile learns a discriminative feature for person re-identification.  ...  However, labeling large-scale training data is very expensive and time-consuming, while large-scale synthetic dataset shows promising value in learning generalizable person re-identification models.  ...  To address this problem, a solution called DomainMix is proposed, for discriminative, domain-invariant, and generalizable person re-identification feature learning.  ... 
arXiv:2011.11953v3 fatcat:tcyb3rzsfva7jkbyghipfg2lfq

Fully Unsupervised Person Re-identification viaSelective Contrastive Learning [article]

Bo Pang, Deming Zhai, Junjun Jiang, Xianming Liu
2021 arXiv   pre-print
Person re-identification (ReID) aims at searching the same identity person among images captured by various cameras.  ...  Representation learning plays a critical role in unsupervised person ReID. In this work, we propose a novel selective contrastive learning framework for unsupervised feature learning.  ...  Especially, for video-based person re-identification, RACE [18] firstly adopt anchor sequences to formulate an anchor graph.  ... 
arXiv:2010.07608v2 fatcat:c4zqhhd2wvex7ijmmvqh667a4m

Dual Distribution Alignment Network for Generalizable Person Re-Identification [article]

Peixian Chen, Pingyang Dai, Jianzhuang Liu, Feng Zheng, Qi Tian, Rongrong Ji
2020 arXiv   pre-print
Domain generalization (DG) serves as a promising solution to handle person Re-Identification (Re-ID), which trains the model using labels from the source domain alone, and then directly adopts the trained  ...  Such an alignment is conducted by dual-level constraints, i.e., the domain-wise adversarial feature learning and the identity-wise similarity enhancement.  ...  RELATED WORK Person Re-identification.  ... 
arXiv:2007.13249v1 fatcat:k2ppkpeqrrhaplmyngevswekpi

A Comprehensive Overview of Person Re-Identification Approaches

Hongbo Wang, Haomin Du, Yue Zhao, Jiming Yan
2020 IEEE Access  
Person re-identification, identifying and tracking pedestrians in cross-domain monitoring systems, is an important technology in the computer vision field and of real significance for the construction  ...  Moreover, this overview summarizes the difficulties and challenges of re-identification and discusses the possible research directions for reference.  ...  re-identification researches mainly start from several perspectives: transfer learning [303] , feature representation extraction [304] , [307] , dictionary learning [302] , clustering [305] , [306  ... 
doi:10.1109/access.2020.2978344 fatcat:amocokxx25btbdeswuqxasox4u
« Previous Showing results 1 — 15 out of 23,177 results