Multiple object tracking based on multi‐task learning with strip attention

Abstract Multiple object tracking (MOT) framework based on bifurcate strategy was usually challenged by data association of different model path, which work for object localisation and appearance embedding independently. By incorporating the re‐identification (re‐ID) as appearance embedding model, m...

Descripción completa

Guardado en:

Detalles Bibliográficos
Autores principales:	Yaoye Song, Peng Zhang, Wei Huang, Yufei Zha, Tao You, Yanning Zhang
Formato:	article
Lenguaje:	EN
Publicado:	Wiley 2021
Materias:	Photography TR1-1050 Computer software QA76.75-76.765
Acceso en línea:	https://doaj.org/article/a098874d1fbd491b82c79120db0433c9
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

id	oai:doaj.org-article:a098874d1fbd491b82c79120db0433c9
record_format	dspace
spelling	oai:doaj.org-article:a098874d1fbd491b82c79120db0433c92021-11-29T03:38:16ZMultiple object tracking based on multi‐task learning with strip attention1751-96671751-965910.1049/ipr2.12327https://doaj.org/article/a098874d1fbd491b82c79120db0433c92021-12-01T00:00:00Zhttps://doi.org/10.1049/ipr2.12327https://doaj.org/toc/1751-9659https://doaj.org/toc/1751-9667Abstract Multiple object tracking (MOT) framework based on bifurcate strategy was usually challenged by data association of different model path, which work for object localisation and appearance embedding independently. By incorporating the re‐identification (re‐ID) as appearance embedding model, more recent studies on task combination of a single network have made a great progress in tracking performance. Unfortunately, the contributive improvement from re‐ID model is hard to balance the accuracy and efficiency for the whole framework. For more effective enhancement of the overall tracking performance, a real‐time detection needs to be taken into consideration with other auxiliary means for MOT modelling. Therefore, in this study, a one‐shot multiple object tracking is proposed based on multi‐task learning to obtain satisfactory performance in both speed and robustness. With updated re‐training strategy for the backbone model of detection, a D2LA network is proposed to achieve more characteristic fine‐grained feature extraction in branching task of pedestrian recognition. Additionally, a strip attention module is also introduced to further strengthen the feature discriminative capability of the tracking framework in occlusion. Experiments on the 2DMOT15, MOT16, MOT17, and MOT20 benchmark data sets have shown a superior performance in comparison to other state‐of‐the‐art tracking approaches.Yaoye SongPeng ZhangWei HuangYufei ZhaTao YouYanning ZhangWileyarticlePhotographyTR1-1050Computer softwareQA76.75-76.765ENIET Image Processing, Vol 15, Iss 14, Pp 3661-3673 (2021)
institution	DOAJ
collection	DOAJ
language	EN
topic	Photography TR1-1050 Computer software QA76.75-76.765
spellingShingle	Photography TR1-1050 Computer software QA76.75-76.765 Yaoye Song Peng Zhang Wei Huang Yufei Zha Tao You Yanning Zhang Multiple object tracking based on multi‐task learning with strip attention
description	Abstract Multiple object tracking (MOT) framework based on bifurcate strategy was usually challenged by data association of different model path, which work for object localisation and appearance embedding independently. By incorporating the re‐identification (re‐ID) as appearance embedding model, more recent studies on task combination of a single network have made a great progress in tracking performance. Unfortunately, the contributive improvement from re‐ID model is hard to balance the accuracy and efficiency for the whole framework. For more effective enhancement of the overall tracking performance, a real‐time detection needs to be taken into consideration with other auxiliary means for MOT modelling. Therefore, in this study, a one‐shot multiple object tracking is proposed based on multi‐task learning to obtain satisfactory performance in both speed and robustness. With updated re‐training strategy for the backbone model of detection, a D2LA network is proposed to achieve more characteristic fine‐grained feature extraction in branching task of pedestrian recognition. Additionally, a strip attention module is also introduced to further strengthen the feature discriminative capability of the tracking framework in occlusion. Experiments on the 2DMOT15, MOT16, MOT17, and MOT20 benchmark data sets have shown a superior performance in comparison to other state‐of‐the‐art tracking approaches.
format	article
author	Yaoye Song Peng Zhang Wei Huang Yufei Zha Tao You Yanning Zhang
author_facet	Yaoye Song Peng Zhang Wei Huang Yufei Zha Tao You Yanning Zhang
author_sort	Yaoye Song
title	Multiple object tracking based on multi‐task learning with strip attention
title_short	Multiple object tracking based on multi‐task learning with strip attention
title_full	Multiple object tracking based on multi‐task learning with strip attention
title_fullStr	Multiple object tracking based on multi‐task learning with strip attention
title_full_unstemmed	Multiple object tracking based on multi‐task learning with strip attention
title_sort	multiple object tracking based on multi‐task learning with strip attention
publisher	Wiley
publishDate	2021
url	https://doaj.org/article/a098874d1fbd491b82c79120db0433c9
work_keys_str_mv	AT yaoyesong multipleobjecttrackingbasedonmultitasklearningwithstripattention AT pengzhang multipleobjecttrackingbasedonmultitasklearningwithstripattention AT weihuang multipleobjecttrackingbasedonmultitasklearningwithstripattention AT yufeizha multipleobjecttrackingbasedonmultitasklearningwithstripattention AT taoyou multipleobjecttrackingbasedonmultitasklearningwithstripattention AT yanningzhang multipleobjecttrackingbasedonmultitasklearningwithstripattention
_version_	1718407663606300672

Multiple object tracking based on multi‐task learning with strip attention

Ejemplares similares