Abstract: In order to improve the efficiency and accuracy of the monitoring personal and vehicles on the substation construction site, a method of non-perceptible interactive based on multiple video analysis technologies is established to replace traditional monitor methods and to build a smart construction site. First, a basic information database of on-site personnel and vehicles in the monitor software system is built to provide a benchmark for video analysis and comparison. Then two-way personnel / vehicle identification drums, panoramic cameras and perimeter dome cameras are deployed at the main entrance and perimeter of the substation. Face recognition, intrusion detection and other video analysis technologies for real-time recognition and analysis of video streams are used to check the attendance of personnel and vehicles and to trigger abnormal behavior alarms. The average multi-target recognition accuracy rate of this research application can reach 95% , which greatly improves the efficiency and accuracy of the entry and exit control of personnel and vehicles on the substation construction site.

Key words: smart construction site, non-perceptibleinteraction, dynamic multi-face recognition technology, license plate recognition technology, behavior monitoring technology, personnel and vehicle monitor