Currency Detection App for Visually impaired
Rahamath Nisha · Zenodo (CERN European Organization for Nuclear Research) · 2023
Apart from the immediate use of Master cards and any form of electronic payment methods, money has been widely used for general exchange due to its usefulness. However, visually impaired people can suffer to know each currency paper separately. This research project aims to develop a mobile application based on the Convolutional Neutral Network using pre-trained representations to permit visually impaired people to see Indian banknotes to real time situations. The elementary methods used in our planned system include pre-image classification, segmentation, histogram equalization, Region of Interest (ROI), and finally template matching using MobileNetV1-224 and Tensor Flow. Visually Impaired are those people who have vision impaired in performing daily activities are in great number. They also face a lot of difficulties in monetary transactions. They are unable to recognize the paper currencies due to similarity of paper texture and size between different categories. This money detector app helps visually impaired patients to recognize and detect money. Using this application blind people can speak and give command to open camera of a smartphone and camera will click picture of the note and tell the user by speech how much the money note is. This Android project uses speech to text conversion to convert the command given by the blind patient. Speech Recognition is a technology that allows users to provide spoken input into the systems. This android application uses text value into speech. For currency detection, this application uses Azure custom vision API using Machine learning classifications technique to detect currency based on images or paper using mobile camera.