Machine Learning Approaches for Banknote Recognition: A Comparative Study of KNN and Random Forest
Keywords: banknote recognition, machine learning, assistive technology, visually impaired
Abstract: Millions of individuals worldwide suffer from reduced or absent vision.It might be challenging for those with visual impairments to discern between different cash denominations and currencies. This work takes a dataset of different currencies, trains the model to recognise the denomination, currency, and orientation, and then provides the user with the information.The dataset consists of 24,826 photos of banknotes in various adaptive settings, including 112 denominations and 17 different currencies. We create a machine learning model for worldwide money recognition using supervised contrastive learning. With the help of this model, banknote images can be embedded compliantly in a range of scenarios.We compare the performance of two machine learning algorithms—K-Nearest Neighbors (KNN) and Random Forest (RF)—in recognizing the banknotes.Our analysis provides insights into the feasibility of each method for practical assistive applications by comparing their relative efficiency in terms of accuracy and computing performance.
Submission Number: 20
Loading