Cargando…

Pixel-Guided Association for Multi-Object Tracking

Propagation and association tasks in Multi-Object Tracking (MOT) play a pivotal role in accurately linking the trajectories of moving objects. Recently, modern deep learning models have been addressing these tasks by introducing fragmented solutions for each different problem such as appearance mode...

Descripción completa

Detalles Bibliográficos
Autores principales: Boragule, Abhijeet, Jang, Hyunsung, Ha, Namkoo, Jeon, Moongu
Formato: Online Artículo Texto
Lenguaje:English
Publicado: MDPI 2022
Materias:
Acceso en línea:https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9692782/
https://www.ncbi.nlm.nih.gov/pubmed/36433519
http://dx.doi.org/10.3390/s22228922
_version_ 1784837355152080896
author Boragule, Abhijeet
Jang, Hyunsung
Ha, Namkoo
Jeon, Moongu
author_facet Boragule, Abhijeet
Jang, Hyunsung
Ha, Namkoo
Jeon, Moongu
author_sort Boragule, Abhijeet
collection PubMed
description Propagation and association tasks in Multi-Object Tracking (MOT) play a pivotal role in accurately linking the trajectories of moving objects. Recently, modern deep learning models have been addressing these tasks by introducing fragmented solutions for each different problem such as appearance modeling, motion modeling, and object associations. To bring unification in the MOT task, we introduce a pixel-guided approach to efficiently build the joint-detection and tracking framework for multi-object tracking. Specifically, the up-sampled multi-scale features from consecutive frames are queued to detect the object locations by using a transformer–decoder, and per-pixel distributions are utilized to compute the association matrix according to object queries. Additionally, we introduce a long-term appearance association on track features to learn the long-term association of tracks against detections to compute the similarity matrix. Finally, a similarity matrix is jointly integrated with the Byte-Tracker resulting in a state-of-the-art MOT performance. The experiments with the standard MOT15 and MOT17 benchmarks show that our approach achieves significant tracking performance.
format Online
Article
Text
id pubmed-9692782
institution National Center for Biotechnology Information
language English
publishDate 2022
publisher MDPI
record_format MEDLINE/PubMed
spelling pubmed-96927822022-11-26 Pixel-Guided Association for Multi-Object Tracking Boragule, Abhijeet Jang, Hyunsung Ha, Namkoo Jeon, Moongu Sensors (Basel) Article Propagation and association tasks in Multi-Object Tracking (MOT) play a pivotal role in accurately linking the trajectories of moving objects. Recently, modern deep learning models have been addressing these tasks by introducing fragmented solutions for each different problem such as appearance modeling, motion modeling, and object associations. To bring unification in the MOT task, we introduce a pixel-guided approach to efficiently build the joint-detection and tracking framework for multi-object tracking. Specifically, the up-sampled multi-scale features from consecutive frames are queued to detect the object locations by using a transformer–decoder, and per-pixel distributions are utilized to compute the association matrix according to object queries. Additionally, we introduce a long-term appearance association on track features to learn the long-term association of tracks against detections to compute the similarity matrix. Finally, a similarity matrix is jointly integrated with the Byte-Tracker resulting in a state-of-the-art MOT performance. The experiments with the standard MOT15 and MOT17 benchmarks show that our approach achieves significant tracking performance. MDPI 2022-11-18 /pmc/articles/PMC9692782/ /pubmed/36433519 http://dx.doi.org/10.3390/s22228922 Text en © 2022 by the authors. https://creativecommons.org/licenses/by/4.0/Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).
spellingShingle Article
Boragule, Abhijeet
Jang, Hyunsung
Ha, Namkoo
Jeon, Moongu
Pixel-Guided Association for Multi-Object Tracking
title Pixel-Guided Association for Multi-Object Tracking
title_full Pixel-Guided Association for Multi-Object Tracking
title_fullStr Pixel-Guided Association for Multi-Object Tracking
title_full_unstemmed Pixel-Guided Association for Multi-Object Tracking
title_short Pixel-Guided Association for Multi-Object Tracking
title_sort pixel-guided association for multi-object tracking
topic Article
url https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9692782/
https://www.ncbi.nlm.nih.gov/pubmed/36433519
http://dx.doi.org/10.3390/s22228922
work_keys_str_mv AT boraguleabhijeet pixelguidedassociationformultiobjecttracking
AT janghyunsung pixelguidedassociationformultiobjecttracking
AT hanamkoo pixelguidedassociationformultiobjecttracking
AT jeonmoongu pixelguidedassociationformultiobjecttracking