Full metadata record
DC FieldValueLanguage
dc.contributorDepartment of Mechanical Engineeringen_US
dc.contributor.advisorHuang, Hailong (AAE)en_US
dc.creatorWang, Xinran-
dc.identifier.urihttps://theses.lib.polyu.edu.hk/handle/200/13011-
dc.languageEnglishen_US
dc.publisherHong Kong Polytechnic Universityen_US
dc.rightsAll rights reserveden_US
dc.titleAbnormal status detection of construction workersen_US
dcterms.abstractThe state of construction workers on construction sites is crucial for their physical health and work efficiency, among other aspects, and needs special attention. Due to long working hours, physical exertion, and harsh working conditions, workers easily experience fatigue during construction. When fatigued, workers tend to smoke to alleviate tiredness, which poses safety hazards such as fire accidents. Other workers may choose to make phone calls to relieve fatigue. These behaviors can cause distraction and safety hazards, threatening the workers’ safety.en_US
dcterms.abstractTherefore, detecting abnormal worker states becomes essential under the above circumstances. In this paper, we employ two parallel methods to comprehensively identify workers’ abnormal conditions. Firstly, we use the deep neural learning network VGG19 to classify the dataset into fatigue and non-fatigue, with fatigue including yawning and blinking; making phone calls and not making phone calls; smoking and not smoking. Then, we set the training, validation, and testing datasets. After deep learning, the training accuracy can reach above 0.95, and the prediction results have high probabilities in their corresponding categories, with excellent performance that can be applied in real-world scenarios. Secondly, we detect workers’ abnormal states in real-time. For fatigue detection, we use the dlib library, the Perclos fatigue algorithm, and 68 facial landmarks to detect yawning and blinking in real-time and perform counting. When the count exceeds a certain degree, a reminder will appear on the screen. For smoking and phone calls, we use the YOLOv5 object detection algorithm, which displays the location and confidence level of cigarettes or phones on the screen. When the confidence level is above 0.9, a reminder appears on the screen. Lastly, we apply the YOLOv7 algorithm for workers’ trajectory tracking and prediction, selecting linear prediction during the forecast. The accuracy is also excellent after calculation.en_US
dcterms.extentix, 87 pages : color illustrationsen_US
dcterms.isPartOfPolyU Electronic Thesesen_US
dcterms.issued2023en_US
dcterms.educationalLevelM.Sc.en_US
dcterms.educationalLevelAll Masteren_US
dcterms.LCSHConstruction workers -- Health and hygieneen_US
dcterms.LCSHFatigueen_US
dcterms.LCSHConstruction industry -- Safety measureen_US
dcterms.LCSHMachine learningen_US
dcterms.LCSHHong Kong Polytechnic University -- Dissertationsen_US
dcterms.accessRightsrestricted accessen_US

Files in This Item:
File Description SizeFormat 
7479.pdfFor All Users (off-campus access for PolyU Staff & Students only)9.04 MBAdobe PDFView/Open


Copyright Undertaking

As a bona fide Library user, I declare that:

  1. I will abide by the rules and legal ordinances governing copyright regarding the use of the Database.
  2. I will use the Database for the purpose of my research or private study only and not for circulation or further reproduction or any other purpose.
  3. I agree to indemnify and hold the University harmless from and against any loss, damage, cost, liability or expenses arising from copyright infringement or unauthorized usage.

By downloading any item(s) listed above, you acknowledge that you have read and understood the copyright undertaking as stated above, and agree to be bound by all of its terms.

Show simple item record

Please use this identifier to cite or link to this item: https://theses.lib.polyu.edu.hk/handle/200/13011