SOS Detection: Distress Signal Recognition

Abstract

In emergency situations where individuals cannot verbally call for help, there is a critical need for a reliable and efficient non-verbal communication system. This project proposes a real-time SOS gesture recognition solution using advanced computer vision techniques to address this need. The system utilizes MediaPipe for precise hand gesture detection and OpenCV for robust image processing to identify a predefined SOS gesture sequence, specifically an open hand followed by a fist. Upon detecting the SOS gesture, the system captures an image of the individual, overlays a timestamp in the bottom-right corner for context, and promptly sends an email alert with the captured image to a predefined recipient using SMTP (Simple Mail Transfer Protocol). This ensures that the alert is dispatched quickly and reliably. The solution is designed to ensure real-time processing and minimal latency, making it highly responsive in critical situations. Additionally, its ease of deployment and adaptability make it suitable for a wide range of applications, including personal safety, medical emergencies, and security systems. By providing a silent and effective method for signaling distress, this system enhances safety and offers peace of mind in various high-risk environments.