Preventing Threats to Women Through Advanced Surveillance

Women’s safety in both public and private environments requires intelligent systems capable of detecting threats and enabling rapid response. This paper presents an AI-based Women Safety System designed for real-time violence detection and automated emergency alert generation. The system integrates a YOLO (You Only Look Once) deep learning model with OpenCV to analyze live and recorded video streams for identifying aggressive or abnormal behavior. Upon detecting violent activity above a predefined confidence threshold, the system captures visual evidence and automatically generates a structured emergency alert containing the incident timestamp, geographic coordinates, readable address, and a Google Maps navigation link. The alert, along with the captured screenshot, is securely transmitted via email to authorized administrative personnel. Developed using Python and the Django framework, the system ensures secure authentication, low latency processing, and reliable performance under varying environmental conditions. Experimental evaluation demonstrates accurate threat detection and timely alert delivery, significantly reducing response time. The proposed solution contributes to proactive surveillance and strengthens intelligent safety infrastructure for women.