MobileNet Outperforms in Detecting Fake Indian Currency: A Performance Evaluation
B. V. Prasanthi, Sajjad Hussain, E. Shalini, Mukesh Prasad · 2025
The rise of fake currency is a growing concern that threatens the economy, making it crucial to develop effective ways to detect it. This research introduces a new method for identifying forged Indian currency using Convolutional Neural Networks (CNNs). We explored the capabilities of selected CNN models, including AlexNet, ResNet, and MobileNet, to classify currency notes as either real or forged. Our dataset featured images of different Indian denominations, which we pre-processed by resizing and augmenting to enhance the model’s learning ability. We evaluated the models through extensive testing based on accuracy and categorical cross entropy loss. Our research showed that MobileNet performed the best, delivering the highest accuracy, making it a strong candidate for real time detection of forged currency. This study emphasizes the effectiveness of CNNs in providing a fast and reliable solution for counterfeit detection, ultimately contributing to the protection of the economy.