Electoral elections are the basis of democratic governance, with an unparalleled importance that shapes the trajectory of a nation. Current paper-based systems continue to be a significant challenge in many countries, bringing problems related to trust, transparency and security. To avoid this many countries have seen electronic voting as a more attractive and efficient alternative. These electronic systems implemented in some countries, where the advantages of their use have been obvious, have also encountered their own set of challenges. Massive manipulations in voting results can happen even from a very small vulnerability. New technologies that are under development such as biometrics, blockchain and machine learning have been implemented to revolutionize the paradigm of electronic voting. Facial recognition is a modality used by us to authenticate voters, while we use algorithms such as Haar Classifier algorithm for the most accurate recognition of identity to guarantee transparent, safe and avoidable voting. We propose an architecture of the face recognition system, database management and machine learning process. A simulated case study is developed to illustrate the functionality and effectiveness of the system based on the 2020 elections in the United States. The system proposed allows users to automate the identity verification process by ensuring a good database management and a secure network that increases efficiency and transparency. Future research is envisioned to drive improvement of this system by implementing real-world scenarios.