A Transformer-Ready Facial Recognition Platform for Locating Missing Individuals with Real-Time Community Alerts

Every year, thousands of people go missing worldwide, creating emotional and operational challenges for families and law enforcement. SafeReturns++ is an AI-driven platform that uses facial recognition—combining YOLOv8-based face detection with ArcFace-style embeddings—to rapidly identify missing individuals from CCTV feeds and citizen-uploaded images or live video. The system supports case registration via a web portal, performs real-time detection, alignment, embedding, and vector search, and automatically notifies families and authorities upon a match. Geo-tagged sightings generate route maps and community alerts. Evaluated on public datasets and a curated missing-person test set, SafeReturns++ delivers a face detection [email protected] of 96.8%, identification accuracy of 98.9% at 0.1% FAR, and sub-150 ms latency on commodity GPUs. These results show that an AI-powered, community-supported workflow can significantly accelerate and scale missing-person investigations.