One important component of preserving the integrity of the economy is the detection of counterfeit banknotes, and this study suggests using a Convolutional Neural Network (CNN) model for this purpose. The model uses a collection of photos of real and counterfeit money notes of different denominations to train, with pre-processing involving levelling pixel values and scaling each image to a specified size. The pre-processed photos are divided into training and validation sets in order to test and train the model. The model uses the confidence score to distinguish between genuine and counterfeit currencies of different denominations. The CNN model technique has improved currency recognition analysis with improved accuracy, speed, efficiency, and minimal complexity. The project’s primary goal is cash detection, which includes currency denomination. The MobileNet model was used to train the CNN and experiment with various cash images. The trained model achieved 99% accuracy, indicating thorough training and no overfitting during training.