Authors: R. Oviyashree, M. Loganathan, S.P. Manoj, A.S, Vasunthra, Dr.R.Punithavathi
Abstract: Every year, forensic investigators and humanitarian groups are overwhelmed by thousands of missing people and unidentified bodies. However, traditional methods of identifica-tion are still very slow and prone to mistakes. AMPUIS, the AI-based missing person and Unidentified Body Identification System, solve this problem with a smart, real-time forensic framework. The system uses a ResNet-10-based Single Shot Multi-Box Detector to find faces, Open-face to extract deep embeddings that don’t change with pose, and a Support Vector Machine classifier to give probability scored identity verification. AMPUIS is built on a secure Flask web architecture that lets law enforcement and NGOs access it based on their roles. It has automated case management, severity-based alerts, and a live forensic dashboard. Experimental results show that this method is more accurate and faster than traditional biometric methods.It is also scalable and can be used all over the world for modern forensic identification.
DOI: http://doi.org/
