Mflorentine/Enhancing-Customer-Service-Delivery-in-Insurance-Companies-Using-CNN-for-Face-Recognition

A CNN‑based face recognition system using MTCNN, VGGFace, SQLite, and Streamlit to improve customer identification and reduce fraud in insurance companies. Achieves 96% accuracy and helped enhance service efficiency in Rwandan insurance workflows. This project is based on the published research article: “Enhancing Customer Service Delivery in Insurance Companies Using Convolutional Neural Network for Face Recognition: Evidence from Rwanda.” Published in International Journal of Innovative Science and Research Technology, 2024. Features: Automated customer identification using facial recognition Fraud prevention through biometric verification Efficient service delivery using AI‑powered identity matching MTCNN for face detection and VGGFace for recognition SQLite databases for secure client data storage Streamlit interface for image upload, webcam input, and results display Can integrate with existing insurance information systems Technologies Used: Python Multi‑Task Cascaded Convolutional Neural Networks (MTCNN) VGGFace model SQLite Database Streamlit Deep Learning / Convolutional Neural Networks Model Performance: The developed system achieves 96% accuracy, performing better than several existing face recognition approaches and demonstrating strong reliability for real‑world insurance workflows in Rwanda. Use case: Insurance companies often struggle with client identification during service delivery, especially when documentation is missing or when fraud attempts occur. This system helps: Identify clients using facial biometrics Improve service efficiency Reduce fraud risks Enhance customer experience