This study presents the development of a model for classifying images of authentic and counterfeit banknotes using supervised deep learning techniques. The dataset utilized for training and evaluation comprised 1,000 images, evenly distributed between 500 authentic and 500 counterfeit banknote images. The model was trained over 50 epochs. A comparative experiment was conducted to evaluate the performance of two algorithms: a Convolutional Neural Network (CNN) and a CNN based on the MobileNet architecture. The results demonstrate that both algorithms exhibited comparable performance across all evaluated metrics. Specifically, the MobileNet-based CNN achieved superior results compared to the conventional CNN model, with accuracy rates of 100% and 98.50%, precision rates of 100% for both models, recall rates of 100% and 97.16%, and F1-scores of 100% and 98.56%, respectively.