AI-Powered Security Camera System for Real-time Accident and Suspicious Activity Detection
Saahil Santosh Barve, Akshay Ganpat Kesarkar, Heet Bharat Gala
Department of Artificial Intelligence and Data Science
K.J. Somaiya Institute of Technology
Sion, Mumbai
Abstract — In an era of heightened security concerns and the growing need for intelligent surveillance systems, this research presents an innovative AI-based security camera system designed to enhance safety and security in various settings amidst growing security concerns. The system leverages the YOLO (You Only Look Once) object identification algorithm for real-time accident and suspicious activity detection, offering a proactive approach to threat mitigation. Based on comprehensive training data, YOLO's versatility in recognizing diverse objects and situations enables ongoing video stream analysis and immediate alerting of issues. Furthermore, Google Firebase is seamlessly integrated for user authentication and data management, ensuring secure and efficient access control. Firebase enhances data storage and retrieval, contributing to system reliability and scalability. The user interface is built using Streamlit, a Python toolkit, offering an intuitive web-based dashboard for users to adjust system settings, review detected incidents, and monitor the security camera system in real time. The system's evaluation efficiently identifies accidents and suspicious activity, with minimal false positives. Firebase integration ensures robust user access control, and the Streamlit interface enhances user interaction. In conclusion, this AI-powered security camera system with real-time incident detection capabilities provides a comprehensive solution for enhancing security and safety across various scenarios. The combination of YOLO, Google Firebase, and Streamlit offers a powerful and user-friendly system, highlighting the potential of AI-driven solutions in the realm of intelligent surveillance.
Keywords—Artificial Intelligence, YOLO, Streamlit, Firebase.
