Paper Title
Bank Locker Security System Using Machine Learning With Face & Liveness Detection

Ensuring the security of transactions is currently one of the biggest challenges facing banking systems. The use of biometric authentication of users attracts huge sums of money from banks around the world due to their convenience and acceptance. Especially in offline environments, where face images from ID documents are matched to digital selfies. In fact, comparisons of selfies with IDs have also been used in some broader programs these days, such as automatic immigration control. The great difficulty of such a process lies in limiting the differences between comparative facial images given their different origins. we propose a novel architecture for cross-domain matching problem based on deep features extracted by two well-referenced Convolutional Neural Networks (CNN). The results obtained from the data collected, called Face Bank, with more than 93% accuracy, indicate the strength of the proposed face-to-face comparison problem and its inclusion in real banking security systems. Keywords - Convolutional Neural Networks (CNN), Face Bank, automatic immigration control, Digital selfies.