Research on blockchain electronic voting system based on face recognition and deep fake face detection

Abstract

To overcome challenges in voter authentication, fraud prevention, and security of election data, this paper suggests a blockchain-based electronic voting system combining the use of facial recognition, deep fake detection, and vote storage to maintain electoral integrity. Implementing facial recognition is done using more sophisticated deep learning models, whereas deepfake detection is achieved by using convolutional networks with the addition of frequency-domain analysis to reduce threats of identity spoofing. Smart contracts are used to store the votes in a blockchain, ensuring their transparency and immutability, and thus decentralized and auditable storage. Moreover, a Dynamic Revoting Mechanism is proposed, which permits the voters to remove and re-vote during the election. This is to ensure that after each vote is cast, the latest one is added to the final tally to avoid cases of duplication of voting and manipulation. Experiment outcomes have shown high voter authentication accuracy (97.36) and high level of deep fake detection and blockchain integration ensures security and transparency in the storage of votes. The proposed framework is much better in terms of integrity and trustworthiness compared to the traditional e-voting systems. It has provided the technical basis of secure, fraud-resistant, and transparent digital voting, though the applications may be used in a smart city digital ecosystem.