Answers are grounded in this paper and cite their sources.
The main construction phase of bridge projects primarily involves high-altitude operations, where the complex site environment and inadequate prior control often lead to safety accidents. An intelligent safety early warning method for high-altitude operations on bridges is designed using deep learning and computer vision techniques, focusing on personnel trajectory prediction. Initially, You Only Look Once (YOLO)v5 is employed to identify the presence of edge guardrails around the high-altitude operation platform, while the Canny edge detection algorithm extracts the platform edges. Dynamic danger zones for high falls are delineated based on the presence or absence of these guardrails. Subsequently, an improved YOLOv5+Deep Simple Online and Realtime Tracking (DeepSORT) multitarget tracking algorithm is used to track construction workers and extract their trajectories. The social long short-term memory (LSTM) algorithm is then utilized to predict the workers’ future movement trajectories. Finally, the ray casting algorithm determines whether these predicted trajectories enter dangerous zones, providing proactive warnings for potential falls. The results demonstrate that this method accurately predicts worker trajectories and provides timely warnings. This approach effectively addresses the limitations of manual supervision in construction safety management, reduces the likelihood of accidents, and offers a novel method for enhancing bridge construction safety management.
No takes yet. Share an insight, caveat, or question.
Tian et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: