Optimizing Currency Classification and Counterfeit Detection with Machine Learning Algorithms
Dnyaneshwari. B. Narale, A. M. Pujar, Komal Rahul Pardeshi · 2025
Currency classification and counterfeit detection are crucial for protecting financial security and combating fraud. Traditional detection methods often rely on manual checks or hardware-based systems, which can be slow, costly, and difficult to scale. This research presents a smart, machine learning-based approach that leverages Convolutional Neural Networks (CNNs), XGBoost, and Random Forest to effectively classify currency denominations and identify counterfeit notes. The primary aim is to develop an intelligent system that combines deep learning with tailored features to enhance the accuracy and reliability of currency verification. The proposed model employs advanced feature extraction techniques such as Canny edge detection, Gray-Level Co-occurrence Matrix (GLCM), color histograms, and Local Binary Patterns (LBP) to capture both texture and structural details from currency images. A meticulously curated dataset featuring high-resolution images of both genuine and counterfeit Indian banknotes, taken under various lighting conditions and angles, is used for training and evaluation. Experimental results indicate that integrating handcrafted features with deep learning significantly boosts classification performance. Additionally, a user-friendly GUI has been created for real-time model evaluations and image-based predictions, making the system applicable to real-world scenarios. This study highlights the potential of hybrid machine learning techniques in automating counterfeit detection with high accuracy, offering a robust, scalable, and cost-effective solution for financial institutions, retail environments, and currency verification systems.