摘要
为应对校园安全管理碎片化与风险预警滞后性问题,提出一种基于多模态人工智能的校园安全风险智能感知与预警系统。系统融合视觉识别、语音分析、文本挖掘、生物传感等多源数据,构建“感知-分析-决策-响应”全链条防控体系。通过跨模态特征融合机制与轻量化边缘计算架构,实现异常行为识别准确率92.3%,预警响应速度较传统安防系统提升4倍以上。在试点学校应用中,校园暴力事件发生率下降35%,心理危机干预时效提升60%,为全域覆盖、全程可控、全员参与的现代化校园安全治理提供技术支撑与实践范式。
To address the fragmentation of campus safety management and the lag in risk early warning,this paper proposes an intel-ligent perception and early warning system for campus safety risks based on multimodal artificial intelligence.The system integrates multi-source data such as visual recognition,speech analysis,text mining,and biosensing to construct a full-chain prevention and control system of"perception-analysis-decision-making-response".Through the cross-modal feature fusion mechanism and the lightweight edge computing architecture,the accuracy rate of abnormal behavior recognition is achieved at 92.3%,and the speed of early warning re-sponse is significantly improved.The efficiency is more than four times higher than that of the traditional security system.In the pilot school application,the incidence of school violence incidents has decreased by 35%,and the timeliness of psychological crisis interven-tion has increased by 60%,providing technical support and practical models for modern campus safety governance that covers the entire area,is fully controllable throughout the process,and involves all staff.
作者
段伟笑
刘娟
殷仁亮
刘俊杰
DUAN Weixiao;LIU Juan;YIN Renliang;LIU Junjie(Xinyang Vocational College Of Art,Xinyang 464000,China)
出处
《通信与信息技术》
2025年第5期110-114,148,共6页
Communication & Information Technology
关键词
多模态人工智能
平安校园
风险预警
智能感知
Multimodal artificial intelligence
Safe campus
Risk early warning
Intelligent perception
