Secure and Hassle-Free EVM Through Deep Learning Based Face Recognition

The primary right of voting in the elections is the fundamental yardstick of a democratic citizen. During the modern era, Electronic Voting machine (EVM) has been introduced which has marked a significant change in the conventional voting system in India replacing the ballot papers and boxes which were used earlier. Previously, the ballot papers used to consume a lot of time, due to the malpractices like booth-capturing and ballot-box stuffing, leading to more disputes and delayed results announcement. In this paper, we propose a EVM system which captures the facial image of a voter through a deep CNN based face recognizer, verifies it with the pre-captured images in the database, the result being positive, assumes the voter is a valid one, asks him to cast his vote for a political party. After voting, the facial alignment of the voter gets deleted from the system, ensuring the voter to vote for only once. This article describes the methodology adopted towards building the system and compares the performance of the face-recognizer (99.1%) against a baseline.