E-Vote Secure: An AI-Powered Electronic Voting Machine Using Face Recognition and Raspberry Pi

Abstract

Traditional paper-based voting systems are plagued by issues of booth capturing, impersonation, ballot stuffing, and manual
counting errors. This paper presents E-VoteSecure, a novel Electronic Voting Machine (EVM) that integrates Artificial
Intelligence-based facial recognition with Raspberry Pi hardware to create a secure, tamper-proof, and transparent democratic
process. The proposed system employs a K-Nearest Neighbors (KNN) machine learning algorithm trained using OpenCV and
Haar Cascade classifiers to perform real-time biometric voter authentication. The web-based application is built on Python
Flask with a SQLite backend featuring SHA-256 password hashing and a one-vote-per-person constraint enforced at the
database level. A multi-tier role-based system accommodates voters, candidates, and election administrators with dedicated
dashboards, approval workflows, and real-time result visualization. Experimental results demonstrate high face recognition
accuracy, effective duplicate vote prevention, and a streamlined end-to-end voting workflow suitable for deployment in local
elections, institutional polls, and remote constituencies.