Sample Problems in Person Re-Identification
Document Type
Syllabus
Publication Date
1-1-2023
Abstract
Person Re-Identification (Re-ID) is a new technology that has emerged in the field of intelligent video analysis in recent years. It belongs to the category of image processing and analysis in complex video environments. Person Re-ID faces the problem of fast and low-cost learning in the case of small samples. In recent years, many research results have emerged regarding the problem of small-sample learning in person Re-ID. New training samples generated by Cycle-GAN can be used to alleviate the problem of data imbalance in pedestrian Re-ID. Person Re-ID technology can be used to obtain customer's behavior trajectory, obtain customer's digital information, help businesses mine more commercial value, and provide customers with customized services. In addition to the application in offline retail solutions, Re-ID can be used to connect online and offline retail scenarios and provide a “one-stop” consumer service experience.
Identifier
85163566224 (Scopus)
ISBN
[9780367512989, 9781000851908]
Publication Title
Intelligent Image and Video Analytics Clustering and Classification Applications
External Full Text Location
https://doi.org/10.1201/9781003053262-9
First Page
303
Last Page
330
Recommended Citation
Han, Hua and Zhou, Mengchu, "Sample Problems in Person Re-Identification" (2023). Faculty Publications. 2118.
https://digitalcommons.njit.edu/fac_pubs/2118