Machine Learning-Based Counterfeit Currency Detection

Counterfeit currency detection is a critical issue posing significant challenges to economies worldwide. Traditional methods relying on manual inspection are prone to error and inefficiency. To enhance detection accuracy, this study explores advanced machine learning techniques, specifically Support Vector Machine (SVM) and Convolutional Neural Network (CNN) models. Existing SVM models struggle with large-scale datasets, and CNN models require extensive training data. There is also a lack of diversity in counterfeit samples and limited use of augmented data in existing datasets. This research compares SVM and CNN models in identifying counterfeit ₹500 Indian banknotes. Using a dataset of genuine and counterfeit notes, features were extracted with a Histogram of Oriented Gradients (HOG) for SVM. ResNet50 architecture was used for CNN. The dataset was augmented to improve robustness. The CNN model achieved training and validation accuracy, while the SVM model also reached high accuracy. These results highlight the potential of both models in providing accurate and automated counterfeit detection, demonstrating their effectiveness and reliability in this application.