Finding Missing Person Using AI
Author: GEENU LAKSHMAN (MCA student), Mr.P.KALYAN CHAKRAVARTHI (Asst.Prof) Department of Artificial Intelligence & Machine Learning, Godavari Global University, Rajahmundry, AP.
Corresponding Author: Geenu Lakshman
(email-id: lakshmangeenu@gmail.com)
ABSTRACT: India faces a critical public safety challenge with 3,51,248 missing persons cases reported in 2024 (NCRB), where conventional investigation methods suffer from 17-day average delays, 28% human matching error, and fragmented records across 16,000+ police stations. Traditional workflows rely on manual FIR documentation, physical photo albums, and labor-intensive CCTV frame analysis, rendering large-scale tracing practically impossible. Digital platforms like Khoya-Paya provide basic case listing but lack automated facial recognition capabilities, forcing investigators to manually browse thousands of images without intelligent matching.
TraceAI proposes a transformative dual-interface AI platform that integrates MediaPipe Face Mesh (468-point facial landmarks, 97.8% detection accuracy) with crowdsourced public participation to revolutionize missing persons investigations. The system architecture comprises two seamlessly integrated components:
1. Secure Admin Portal (Streamlit/Home.py): Police and NGOs register missing persons cases with automated facial landmark extraction. 468-point coordinate vectors are serialized as JSON and stored in centralized SQLite database alongside case metadata (name, age, location, status).
2. Mobile-Responsive Public Interface (mobile_app.py): Citizens submit real-time sighting photos through intuitive photo upload + location form. Anonymous processing ensures privacy while AI immediately computes matches against centralized database.
3. Core Innovation: Real-time Euclidean distance matching delivers top-5 similarity rankings in 823ms across 10K records, achieving 92.1% precision at 0.6 threshold. Performance benchmarks confirm 187ms/image processing, 98.7% face detection success, and 42ms database queries.
KEYWORDS: Facial Recognition, Missing Persons Investigation, MediaPipe Face Mesh, Streamlit Framework, Crowdsourced Public Safety, SQLite Database, Euclidean Distance Matching, Real-time AI Processing
