Face Detection using Deep Learning to ensure a Coercion Resistant Blockchain-based Electronic Voting
Abstract
An election is a powerful tool of any democratic country, which allows every citizen to exercise their right to vote. Though in-person voting is the most widely used medium for a citizen to vote, circumstances like a pandemic, natural disasters, or other location-based issues could deter an individual from doing so. This work proposes an architecture for a mobile-based internet voting application that could be downloaded on a smartphone, allowing citizens to vote remotely. Existing web-based online voting systems are complicated for novice users, requiring a complex key management process and a lack of coercion resistance voting system. The proposed work is a mobile application, which ensures more population coverage. The work also suggests a key management system for regional election officers, thus freeing novice users from the complications of key management. A face detector is proposed to provide coercion resistance in this online voting system. Face detection uses a deep learning-based multi-task cascaded convolutional neural network (MTCNN). The proposed model has also incorporated multi-factor authentication, blockchain technology, and asymmetric encryption standards to ensure security features required in a voting system while providing a hassle-free voting experience to the voter.
