COUNTERFEIT CURRENCY DETECTION USING CNN AND IMAGE PROCESSING TECHNIQUES

Counterfeit currency is a constant threat to the economy, and innovative and automateddetection methods for preventing loss and restoring economic integrity are urgently calledfor. This paper proposes an approach based on CNN, a neural network powerful in imageclassification, for detecting counterfeit Indian currency notes. To train and test a CNN model,a dataset of Indian currency notes is taken from Kaggle, which contains images of real andcounterfeit currency. Data preprocessing and augmentation such as rescaling and horizontalflipping were used to enhance the model generalization due to the small size of the dataset.Based on feature extraction, the CNN model has used several convolutional and poolinglayers to extract high-level patterns that discriminate real from fake notes. The experimentsproduced an accuracy of 84.11% percent; thus, validation has shown its great promise forapplications of practical counterfeit detection. Precision, recall, and F1-score were alsocomputed to fully evaluate the model's performance, confirming deep learning can providean efficient and automatable way toward currency verification.