AN INTELLIGENT MACHINE LEARNING APPROACH FOR MISSING CHILD IDENTIFICATION USING MULTI-CLASS SVM

Abstract

The increasing number of missing children worldwide has become a significant social and security concern, requiring efficient and intelligent identification systems. Traditional methods such as manual investigation, public notices, and media broadcasts are often time-consuming and less effective when dealing with large volumes of data. This paper proposes an intelligent missing child identification system that combines Deep Learningbased facial feature extraction with Multi-Class Support Vector Machine (SVM) classification. Facial images collected from CCTV surveillance, authorized databases, and user uploads undergo preprocessing techniques including face detection, normalization, resizing, and noise removal. The deep learning model extracts discriminative facial features, while the Multi-Class SVM accurately classifies and identifies children by matching extracted features with stored records. The proposed framework is designed to handle variations in illumination, facial expressions, pose, and partial occlusion, making it suitable for realtime surveillance environments. By integrating automated face recognition with CCTV monitoring and database management, the system improves identification accuracy, reduces false detections, and supports faster response by law enforcement and child welfare agencies.