Summary

A mobile-based system designed to automate the verification and validation of Indonesian E-KTP (identity cards), improving the efficiency and accuracy of manual identity checking processes. This was my thesis project, developed as part of research work combining mobile development, machine learning, and backend engineering to build an end-to-end verification pipeline.

Impact

I developed a custom Convolutional Neural Network (CNN) model for E-KTP classification and verification, achieving 80% accuracy and 87.5% precision in detecting valid identity card images. The model was deployed using a Flask-based REST API to enable server-side inference and seamless integration with mobile clients. The system was fully integrated with Supabase and PostgreSQL for secure authentication and structured data storage. I also built the Flutter mobile application that communicates with the backend, forming a complete end-to- end verification workflow.

What I Learned

Through this project, I gained deep experience in machine learning model development, API deployment, and full-stack system integration across mobile, backend, and AI components.

Screenshots