Current practices are limited to effectively identify unsafe conditions due to a lack of integration between internal traffic control plans (ITCP) that can guide safe activities in construction worksites and safety monitoring systems. To address the limitation, this study proposed the novel concept of a safety monitoring system by leveraging unmanned aircraft systems (UAS), game engine-based ITCP, and deep learning. In this study, workers and equipment were automatically recognized through object detection from aerial images. Through the case study, this proposed concept was validated to monitor the unsafe activities of workers by four rules established. While the limitations of this study were documented, such as the low number of aerial images to train deep learning model, the low performance of object detection, and the errors that occurred when augmenting the detection results in the game engine, this study also emphasized the potential of the proposed digital ITCP-based safety monitoring system.

由于内部交通控制计划（ITCP）之间缺乏集成，无法指导施工现场和安全监控系统的安全活动，因此目前的做法仅限于有效识别不安全状况。为了解决这一局限性，本研究通过利用无人机系统（UAS），基于游戏引擎的ITCP和深度学习提出了一种安全监视系统的新颖概念。在这项研究中，通过从航空影像中进行物体检测，可以自动识别工人和设备。通过案例研究，通过建立的四个规则验证了此提议的概念以监视工人的不安全活动。尽管记录了这项研究的局限性，例如训练深度学习模型的航拍图像数量少，目标检测性能低，