The original paper is in English. Non-English content has been machine-translated and may contain typographical errors or mistranslations. ex. Some numerals are expressed as "XNUMX".
Copyrights notice
The original paper is in English. Non-English content has been machine-translated and may contain typographical errors or mistranslations. Copyrights notice
Un suivi automatique et continu des objets impliqués dans un projet de construction est nécessaire pour des tâches telles que l'évaluation de la productivité, la reconnaissance des comportements dangereux et le suivi des progrès. De nombreuses approches de suivi basées sur la vision par ordinateur ont été étudiées et testées avec succès sur des chantiers de construction ; cependant, leurs applications pratiques sont entravées par la précision du suivi limitée par la nature dynamique et complexe des chantiers de construction (c'est-à-dire encombrement d'arrière-plan, occlusion, échelle et pose variables). Pour obtenir de meilleures performances de suivi, une nouvelle approche de suivi basée sur l'apprentissage en profondeur appelée réseaux neuronaux convolutifs multi-domaines (MD-CNN) est proposée et étudiée. L'approche proposée comprend deux étapes clés : 1) représentation multi-domaines de l'apprentissage ; et 2) suivi visuel en ligne. Pour évaluer l'efficacité et la faisabilité de cette approche, elle est appliquée à un projet de métro à Wuhan en Chine, et les résultats démontrent de bonnes performances de suivi dans des scénarios de construction avec un contexte complexe. L'erreur de distance moyenne et la mesure F pour le MDNet sont respectivement de 7.64 pixels et 67. Les résultats démontrent que l'approche proposée peut être utilisée par les gestionnaires de chantier pour surveiller et suivre les travailleurs afin de prévenir les risques sur les chantiers de construction.
Wen LIU
CCCC Second Harbor Engineering Co., Ltd.
Yixiao SHAO
Huazhong University of Science and Technology
Shihong ZHAI
CCCC Second Harbor Engineering Co., Ltd.
Zhao YANG
CCCC Second Harbor Engineering Co., Ltd.
Peishuai CHEN
CCCC Second Harbor Engineering Co., Ltd.
The copyright of the original papers published on this site belongs to IEICE. Unauthorized use of the original or translated papers is prohibited. See IEICE Provisions on Copyright for details.
Copier
Wen LIU, Yixiao SHAO, Shihong ZHAI, Zhao YANG, Peishuai CHEN, "Computer Vision-Based Tracking of Workers in Construction Sites Based on MDNet" in IEICE TRANSACTIONS on Information,
vol. E106-D, no. 5, pp. 653-661, May 2023, doi: 10.1587/transinf.2022DLP0045.
Abstract: Automatic continuous tracking of objects involved in a construction project is required for such tasks as productivity assessment, unsafe behavior recognition, and progress monitoring. Many computer-vision-based tracking approaches have been investigated and successfully tested on construction sites; however, their practical applications are hindered by the tracking accuracy limited by the dynamic, complex nature of construction sites (i.e. clutter with background, occlusion, varying scale and pose). To achieve better tracking performance, a novel deep-learning-based tracking approach called the Multi-Domain Convolutional Neural Networks (MD-CNN) is proposed and investigated. The proposed approach consists of two key stages: 1) multi-domain representation of learning; and 2) online visual tracking. To evaluate the effectiveness and feasibility of this approach, it is applied to a metro project in Wuhan China, and the results demonstrate good tracking performance in construction scenarios with complex background. The average distance error and F-measure for the MDNet are 7.64 pixels and 67, respectively. The results demonstrate that the proposed approach can be used by site managers to monitor and track workers for hazard prevention in construction sites.
URL: https://global.ieice.org/en_transactions/information/10.1587/transinf.2022DLP0045/_p
Copier
@ARTICLE{e106-d_5_653,
author={Wen LIU, Yixiao SHAO, Shihong ZHAI, Zhao YANG, Peishuai CHEN, },
journal={IEICE TRANSACTIONS on Information},
title={Computer Vision-Based Tracking of Workers in Construction Sites Based on MDNet},
year={2023},
volume={E106-D},
number={5},
pages={653-661},
abstract={Automatic continuous tracking of objects involved in a construction project is required for such tasks as productivity assessment, unsafe behavior recognition, and progress monitoring. Many computer-vision-based tracking approaches have been investigated and successfully tested on construction sites; however, their practical applications are hindered by the tracking accuracy limited by the dynamic, complex nature of construction sites (i.e. clutter with background, occlusion, varying scale and pose). To achieve better tracking performance, a novel deep-learning-based tracking approach called the Multi-Domain Convolutional Neural Networks (MD-CNN) is proposed and investigated. The proposed approach consists of two key stages: 1) multi-domain representation of learning; and 2) online visual tracking. To evaluate the effectiveness and feasibility of this approach, it is applied to a metro project in Wuhan China, and the results demonstrate good tracking performance in construction scenarios with complex background. The average distance error and F-measure for the MDNet are 7.64 pixels and 67, respectively. The results demonstrate that the proposed approach can be used by site managers to monitor and track workers for hazard prevention in construction sites.},
keywords={},
doi={10.1587/transinf.2022DLP0045},
ISSN={1745-1361},
month={May},}
Copier
TY - JOUR
TI - Computer Vision-Based Tracking of Workers in Construction Sites Based on MDNet
T2 - IEICE TRANSACTIONS on Information
SP - 653
EP - 661
AU - Wen LIU
AU - Yixiao SHAO
AU - Shihong ZHAI
AU - Zhao YANG
AU - Peishuai CHEN
PY - 2023
DO - 10.1587/transinf.2022DLP0045
JO - IEICE TRANSACTIONS on Information
SN - 1745-1361
VL - E106-D
IS - 5
JA - IEICE TRANSACTIONS on Information
Y1 - May 2023
AB - Automatic continuous tracking of objects involved in a construction project is required for such tasks as productivity assessment, unsafe behavior recognition, and progress monitoring. Many computer-vision-based tracking approaches have been investigated and successfully tested on construction sites; however, their practical applications are hindered by the tracking accuracy limited by the dynamic, complex nature of construction sites (i.e. clutter with background, occlusion, varying scale and pose). To achieve better tracking performance, a novel deep-learning-based tracking approach called the Multi-Domain Convolutional Neural Networks (MD-CNN) is proposed and investigated. The proposed approach consists of two key stages: 1) multi-domain representation of learning; and 2) online visual tracking. To evaluate the effectiveness and feasibility of this approach, it is applied to a metro project in Wuhan China, and the results demonstrate good tracking performance in construction scenarios with complex background. The average distance error and F-measure for the MDNet are 7.64 pixels and 67, respectively. The results demonstrate that the proposed approach can be used by site managers to monitor and track workers for hazard prevention in construction sites.
ER -