Campus intrusion false alarm suppression method based on multi-modal perception and related equipment

Abstract

This application provides a method and related equipment for suppressing false alarms of campus intrusion based on multimodal perception, belonging to the field of security monitoring technology. The method includes: receiving raw sensor data; transforming the coordinates of the raw sensor data to the same spatial coordinate system to obtain spatiotemporally aligned multi-source perception data; substituting the collected environmental interference data into preset calculation rules to calculate reliability weight coefficients and obtain a dynamic reliability weight matrix; extracting features from the multi-source perception data to obtain a feature vector set; inputting the feature vector set into the first channel of a dual-channel fusion model for splicing and multi-head attention processing; inputting the dynamic reliability weight matrix into the second channel of the dual-channel fusion model as an attention coefficient correction factor; integrating the dual-channel information to obtain an intrusion detection confidence level; and making hierarchical alarm decisions based on the intrusion detection confidence level to obtain alarm information. This application can achieve high-precision, low-false-alarm, and adaptive campus perimeter intrusion detection.

Images