Finding Missing Persons using AI: A Face Recognition Method

The worldwide law enforcement agencies have to deal with a great number of challenges in finding missing people, especially when they have limited computational resources and infrastructure constraints. This paper introduces a resource, friendly facial identification framework that is specially designed for constrained environments. Our method makes use of MediaPipe's 478, point facial mesh to create 1, 434, dimensional geometric feature vectors, hence no GPU hardware or large training datasets are required. To speed up similarity searches we have implemented Ball, Tree spatial indexing, which is 4. 8 faster in retrieving results than linear search methods. The system architecture offers two access points: a safe law enforcement interface and a citizen reporting portal, thus facilitating collaborative search efforts. The experiments with a simulated dataset of 600 different individuals show 92. 3.