Advances in Constructional Engineering

As the rapid advancement of smart construction sites progresses, AI inspection systems have become a crucial tool for ensuring construction safety and efficiency. However, high false alarm rates not only affect system reliability but also increase the cost of manual verification, limiting their widespread application. This paper focuses on the issue of false alarm rates in AI inspection systems for smart construction sites, analyzing its causes. By integrating multi-source data fusion, optimizing deep learning models, and dynamically adjusting thresholds, we aim to effectively reduce false alarm rates, enhance system accuracy and stability, and provide technical support for the safety management of smart construction sites.

Smart construction site; AI inspection; False alarm rate; Deep learning; Data fusion