Next-Gen ATM Security: Combining Facial Recognition and User Consent via Deep Learning

Aim: The aim of this research is to improve the security of ATMs by creating ATM models that make use of Convolutional Neural Networks (CNN) and Recurrent Neural Networks (RNN) for facial recognition of users and user approval.Materials and Methods: The study comprises two groups: Group 1 has a CNN-only model with a sample size of 26 samples, and Group 2 uses a hybrid CNN-RNN model with the same sample size of 26. The statistical analysis of the study was conducted with a G Power of 80%, a significance level of 0.05%, and a confidence interval of 95%. Results: The findings indicate that the hybrid CNN-RNN model outperforms the CNN-only model by a significant margin. The accuracy of the hybrid model was between 91.8% and 98.7%, while that of the CNN-only model was between 85.3% and 92.4%.The maximum accuracy found was 98.7% at p-value 0.0480, with p = 0.005. But the dataset for this research was only 26 samples per group, which is a concern regarding the model's generalizability to larger, real-world datasets. More research is required to evaluate how this model would scale and perform on larger datasets to establish its wider applicability. Conclusion: The results indicate that the hybrid CNN-RNN model performs better than the CNN-only system in facial recognition and enhances ATM security. Yet, scalability and generalizability of the model would require further investigation with larger datasets.