LRCN-Based Surveillance System

A Surveillance and Alert System is designed to detect violent activities and fire spread within a surveillance environment, promptly alerting the concerned authority. Today, CCTV surveillance is the most fundamental and impactful security feature a premise can have. Being a well-known method of preventing and identifying undesirable activities, CCTV systems are commonly deployed in various premises. For instance, a college campus may have CCTV installed in diverse structures, such as dormitories, classrooms, canteens, sports complexes, auditoriums, etc. However, manually searching through recorded video after an incident occurs is a time-consuming process. The aim of this project is to develop a Long-term Recurrent Convolutional Network (LRCN) system for environments like academic campuses to monitor CCTV footage and detect violent activities, such as fighting, and fire incidents. The system will create alerts through calls and messages via email, including the video of the detected activity, which will be sent to the respective authority. Keywords—Violence Detection, Alert Generation, LRCN, Fire detection, Non-violence