Anomaly Detection & Classification Using Machine Learning for Campus Surveillance
Shakshi Richhariya · 2024
This study introduces an advanced system for detecting and classifying unusual activities using machine learning, aimed at improving campus security. Instead of relying on people to constantly watch surveillance footage, which can lead to mistakes, this system uses machine learning to automatically identify and sort unusual behaviors captured by cameras. It combines Convolutional Neural Networks (CNNs) for identifying specific features and patterns in the video with Long Short-Term Memory networks (LSTMs) for understanding the sequence of events over time. This method simplifies the process of spotting and categorizing odd or suspicious activities, making it more accurate and reducing the chances of overlooking something important. The process involves preparing the video data, using CNNs to pick out important details, detecting anomalies over time with LSTMs, and then classifying them into different categories. This system is anticipated to significantly enhance the effectiveness in identifying and classifying anomalies, thereby contributing to a safer campus environment by swiftly pinpointing and addressing potential security concerns.