Pakistani Currency Recognition and Classification for Visually Impaired People Using Convolutional Neural Network
Asra Nasir, Sidra Zafar, Zaeem Nasir · 2024
Contemporary advancements aim to enhance the quality of life across various domains, including healthcare and education, through the integration of innovative technologies. In healthcare, computer vision technology is emerging as a valuable tool to assist visually impaired individuals in their daily activities, fostering greater independence and reducing reliance on others. Our focus is on aiding visually impaired individuals in distinguishing between different denominations of Pakistani banknotes. Employing Deep Learning techniques, we propose a multiclass classification model to categorize banknotes. This system will generate auditory cues corresponding to recognized banknote images, facilitating comprehension of their denominations by visually impaired individuals. Our study utilizes a dataset comprising 4,713 image samples of seven Pakistani currency denominations. We will implement a Convolutional Neural Network (CNN) VGG-16 architecture using this database to accurately differentiate between different denominations of Pakistani banknotes. This methodology bridges the gap between the generalization capabilities of existing models and the unique characteristics of Pakistani banknotes, promising effective banknote denomination classification for visually impaired individuals.