Person re-identification using masked keypoints

Diego Reyes, John Atkinson

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

In this work, a method for person re-identification from surveillance videos is proposed. In this approach, person detection is based on moving objects from sequences of images, and on incorporating a feature extraction technique that can distinguish distinct persons according to their physical appearance by using masked images that reduce noise from the background. Our approach uses keypoints to build an image’s descriptor so that the best discriminative keypoints can be identified between different persons. Experiments using our masked re-identification method show significant improvements in the recognition rate when masked frames are used to reduce noise of the second plane.

Original languageEnglish
Title of host publicationRecent Trends and Future Technology in Applied Intelligence - 31st International Conference on Industrial Engineering and Other Applications of Applied Intelligent Systems, IEA/AIE 2018, Proceedings
EditorsOtmane Ait Mohamed, Malek Mouhoub, Samira Sadaoui, Moonis Ali
PublisherSpringer Verlag
Pages45-56
Number of pages12
ISBN (Print)9783319920573
DOIs
StatePublished - 2018
Externally publishedYes
Event31st International Conference on Industrial, Engineering and Other Applications of Applied Intelligent Systems IEA/AIE 2018 - Montreal, Canada
Duration: 25 Jun 201828 Jun 2018

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume10868 LNAI
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference31st International Conference on Industrial, Engineering and Other Applications of Applied Intelligent Systems IEA/AIE 2018
Country/TerritoryCanada
CityMontreal
Period25/06/1828/06/18

Fingerprint

Dive into the research topics of 'Person re-identification using masked keypoints'. Together they form a unique fingerprint.

Cite this