Abstract: Identifying and classifying counterfeit and genuine currency has become essential to protecting economies around the world.Advanced systems to efficient Counterfeit cash has been a major issue in India; according to a 2016 estimate, there was approximately ₹1,000 crore worth of counterfeit currency in circulation. Traditional methods for counterfeit detection have relied heavily on physical characteristics such as watermarks, security threads, and ultraviolet detection, but these techniques have proven insufficient against the sophistication of contemporary counterfeiters.In order to classify currencies and detect counterfeits, this research makes use of deep learning, more especially DenseNet.Using a dataset of 3753 images of Indian Rupee currency notes sourced from Kaggle, the model was trained to classify currency into eight distinct classes, distinguishing between real and fake notes. The model's performance was enhanced by using the DenseNet architecture, which is renowned for its effective feature reuse and increased accuracy. A web-based user interface was developed to allow users to upload currency images and receive instant feedback on the authenticity of the notes.In order to tackle the problem of counterfeit currency, this approach offers a scalable and easily accessible solution.

Keywords: Fake currency detection, Real currency classification, Deep learning, DenseNet, Image classification, currency security, Indian Rupee, Neural networks.

| DOI: 10.17148/IARJSET.2025.12441