Talha JalalWajid ALiMalikUsman EjazMuhammad Hamza Siddique2025-12-172025-12-172024https://escholar.umt.edu.pk/handle/123456789/16734The issue of safety is multifaceted at construction sites, needing strong measures for risk mitigation and lightening loads on personnel and assets. In this regard, the confluence of advanced technologies, such as computer vision and artificial intelligence, has increasingly been applied to bring novel solutions in the augmentation of safety protocols. With the advancement in object detection models, mainly based on the YOLO (You Only Look Once) architecture, that has been the strong way toward the automation of the detection and localization of safety-critical elements within the construction environment. This research introduces an intricately fine-tuned YOLOv8 model, finely tuned for detection of construction safety. Therefore, this work introduces an advanced approach to training and techniques for fine-tuning a YOLOv8 model capable of detecting a wide range of safety entities, viz. hardhats, masks, safety vests, personnel, machinery, and vehicles in static images and video feeds. The YOLOv8 model is real-time and highly precise and accurate in inference, aiming to set new benchmarks in safety at the construction site—that of the rare accident, satisfied safety regulators, and continuous good health of its workers and assets. This article, therefore, presents the multifaceted capabilities and disruptive potential that shape the YOLOv8 model in the augmentation of safety frameworks and risk mitigation across the dynamic and challenging landscapes of the construction environment. This research is fulfilling primarily sustainable development goal number 3 i.e. good health and wellbeing and secondarily sustainable development goal number 9 i.e. industry, innovation and infrastructure. Index TermsenThesis