Real-time intelligent fire identification and early warning method based on campus surveillance video

Abstract

In recent years, campus fire incidents have become frequent, significantly impacting campus safety. Meanwhile, intelligent fire detection and early warning methods have been proposed and applied in forest fire prevention and urban building fire prevention. This paper addresses the issue of campus fires by first proposing a fire detection model based on object detection algorithms, trained with fire sample data. The trained model achieved an accuracy of 94% and a recall rate of 92%. Next, executable files were created to connect campus video data with the program, facilitating intelligent and convenient campus fire early warning through the collaborative work of monitoring devices, displays, and servers. Finally, different fire prevention measures were proposed for different areas of the campus.