Deep learning-based person re-identification methods: A survey and outlook of recent works

Ming, Zhangqiang; Zhu, Min; Wang, Xiangkun; Zhu, Jiamin; Cheng, Junlong; Gao, Chengrui; Yang, Yong; Wei, Xiaoyong

doi:10.1016/j.imavis.2022.104394

Cited by 77 publications

(25 citation statements)

References 31 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Ming et al conducted a study on identifying a person having a same identity from several cameras [16]. Because they focused on image processing using a deep learningbased approach, the objective of their study differed from that of our study.…”

Section: Related Workmentioning

confidence: 96%

Internet News User Analysis Using Deep Learning and Similarity Comparison

Choi

2022

Electronics

View full text Add to dashboard Cite

Nowadays, many Korean users read news from portal sites like Naver and Daum. Users can comment on news articles on such sites, and some try to influence public opinion through their comments. Therefore, news users need to be analyzed. This study proposes a deep learning method to classify each user’s political stance. Further, a method is developed to evaluate how many similar comments each user writes, and another method is developed to evaluate the similarity of a user’s comments with other users’ comments. We collect approximately 2.68 million comments from hundreds of thousands of political news articles in April 2017. First, for the top 100 news users, we classify each user’s political stance with 92.3% accuracy by using only 20% of data for deep learning training. Second, an evaluation of how many similar comments each user writes reveals that six users score more than 80 points. Third, an evaluation of the similarity of each user’s comments to other users’ comments reveals that 10 users score more than 80 points. Thus, based on this study, it is possible to detect malicious commenters, thereby enhancing comment systems used in news portal websites.

show abstract

Section: Related Workmentioning

confidence: 96%

Internet News User Analysis Using Deep Learning and Similarity Comparison

Choi

2022

Electronics

View full text Add to dashboard Cite

show abstract

“…In [20], Ming et al grouped deep learning-based person Re-ID methods into four categories. These categories are: 1. depth metric learning, 2. local feature learning, 3. generative adversarial learning, and 4. sequences feature learning.…”

Section: Background and Related Workmentioning

confidence: 99%

“…This is done through a base network which is commonly an ImageNet pre-trained network, in this case, ResNet50. However, although this approach is more computation friendly, it is difficult for this method to capture the detailed information about the pedestrian [20]. On the other hand, methodologies for extracting local features are various.…”

Section: Feature Learningmentioning

confidence: 99%

See 1 more Smart Citation

Modified Centroid Triplet Loss for Person Re-Identification

Alnissany

Dayoub

2022

Preprint

View full text Add to dashboard Cite

Person Re-identification (Reid) is the process of matching target individuals to their images within different images or videos captured from a variety of angles or cameras. This is a critical task for surveillance applications, in particular, these applications that operate in large environments such as malls and airports. Recent studies use data-driven approaches to tackle this problem. This work continues on this path by presenting a modification of a previously defined loss, the centroid triplet loss (ctl). The proposed loss, modified centroid triplet loss (mctl), emphasizes more on the interclass distance. It is divided to two parts, one penalize for interclass distance and second penalize for intraclass distance. As a result of this adjustment, a new state-of-the-art results of 98.4% rank1, 98.63% mAP on market-1501, and 96.8% rank1, 97.3% mAP on DukeMTMC-Reid were achieved.

show abstract

“…As such, several works [10,11] have focused on learning visual representations across these information sources. However, existing works have mainly focused on methods to eliminate the domain discrepancies and align the different information sources, while investigation on the use of different auxiliary information sources that can be used to provide complementary information or methods to combine such auxiliary information has not hitherto investigated in depth [12].…”

Section: Introductionmentioning

confidence: 99%

Using Auxiliary Information for Person Re-Identification -- A Tutorial Overview

Fernando¹,

Fookes²,

Sridharan³

et al. 2022

Preprint

View full text Add to dashboard Cite

Person re-identification (re-id) is a pivotal task within an intelligent surveillance pipeline and there exist numerous re-id frameworks that achieve satisfactory performance in challenging benchmarks. However, these systems struggle to generate acceptable results when there are significant differences between the camera views, illumination conditions, or occlusions. This result can be attributed to the deficiency that exists within many recently proposed re-id pipelines where they are predominately driven by appearance-based features and little attention is paid to other auxiliary information that could aid the re-id. In this paper, we systematically review the current State-Of-The-Art (SOTA) methods in both uni-modal and multimodal person re-id. Extending beyond a conceptual framework, we illustrate how the existing SOTA methods can be extended to support these additional auxiliary information and quantitatively evaluate the utility of such auxiliary feature information, ranging from logos printed on the objects carried by the subject or printed on the clothes worn by the subject, through to his or her behavioural trajectories. To the best of our knowledge, this is the first work that explores the fusion of multiple information to generate a more discriminant person descriptor and the principal aim of this paper is to

show abstract

Deep learning-based person re-identification methods: A survey and outlook of recent works

Cited by 77 publications

References 31 publications

Internet News User Analysis Using Deep Learning and Similarity Comparison

Internet News User Analysis Using Deep Learning and Similarity Comparison

Modified Centroid Triplet Loss for Person Re-Identification

Using Auxiliary Information for Person Re-Identification -- A Tutorial Overview

Contact Info

Product

Resources

About