Electronic voting was more practical and less expensive than the conventional technique of using paper ballots, but it was also perceived as being less reliable because anyone with physical access to the voting system could interfere with it and alter the outcome of the election. Additionally, it seriously compromises both voting rights and openness. Electronic voting is more convenient and economical than the traditional technique, which involves using paper ballots. Due to the possibility of interference and result manipulation by anyone with physical access to the voting equipment, doubts exist about the validity of the system. The goal of this research is to provide a three-step electronic voting mechanism that uses facial recognition, one-time password (OTP), and voter identification numbers. Convolutional Neural Network (CNN) and Haar Cascade have both been employed for facial recognition. To ensure quicker computation, sections of the dataset are implemented. With this strategy, proxy voting is not only impossible but the voting system’s security is also improved. Extensive analysis is done employing databases of various sizes, along with comparisons with other approaches, to assess the performance of the proposed system.