A Campus Bullying Online Detection System Based on Wi-Fi Sensing and Deep Learning

Campus bullying is a significant global issue that profoundly impacts the mental and physical health of victims. To address this challenge, this paper proposes an online campus bullying detection system that leverages Wi-Fi sensing and deep learning technologies. The system utilizes the PicoScenes platform to collect Channel State Information (CSI) and employs an Attention-based Bidirectional Long Short-Term Memory network (ABLSTM) for behavior recognition. It innovatively designs dedicated software scripts and a user-friendly interface to achieve automatic reception, forwarding, parsing, and preprocessing of CSI data. Additionally, the system supports the training of predictive models and enables online behavior detection. The experimental results indicate that the system's modules collaborate seamlessly, with smooth script execution and a positive human-computer interaction experience. It is capable of online and efficient detection of bullying behaviors, achieving an accuracy rate of no less than 80%.