Fake Indian Paper Currency Detection Using Deep Learning Techniques

Banknote authentication is maintaining the integrity of financial systems worldwide. As counterfeit currency continues to pose a threat, the integration of deep learning algorithms offers a promising solution for effective and efficient banknote verification. This research paper presents a deep learning model for authenticating Indian currency. Gradient-weighted Class Activation Mapping was used in convolutional neural network for finding region of interest. Publicly available datasets are used for genuine and counterfeit banknote images. The performance of the model was evaluated on the basis of accuracy, precision, and recall. Our findings showcase the strengths and limitations of proposed model, shedding light on their effectiveness in different scenarios. The study also addresses challenges encountered, proposes potential improvements, and outlines future directions for enhancing banknote authentication through advanced deep learning methodologies and use of GAN. The proposed model contributes to the ongoing efforts to fortify financial systems against counterfeit threats and underscores the pivotal role of deep learning in authentication of Indian paper currency.