Intelligent Malaysian banknote recognition system for blind people

Tan, Guan Seng (2020) Intelligent Malaysian banknote recognition system for blind people. Project Report. Universiti Teknikal Malaysia Melaka, Melaka, Malaysia. (Submitted)

Preview

Text (Full Text) Intelligent Malaysian banknote recognition system for blind people.pdf - Submitted Version Download (2MB) | Preview

Abstract

Blind people faced a lot of difficulties in order to interact with the environment because most of the information encoded is visual. One of the serious problems they faced is identifying and recognizing the different currency amount of banknote. This paper presents the development of a system with the aim to classify the Malaysian banknotes for blind people so that allows blind people to handle the banknotes in their daily life independently. The objectives of this project are to study Tensorflow deep learning platform for Malaysian Banknote classification, to develop a Malaysian Banknote classification mobile application using image classification with Tensorflow deep learning platform as well as to analysis and identify the classification performance (accuracy and inferencing time) of developed application against different brightness (dark and bright). The development of this system involves the use of Tensorflow to train a newly classification model for Malaysian Banknote and android studio software for application. For the hardware part, a smartphone as an application will be used for blind people in classifying Malaysian banknote. A result of 80% and 83.33% accuracies with an average time of 486.10ms and 488.69ms for dark area and bright area respectively is achieved. In the end of the project, an android application which can be used for banknote classification is successfully developed with high accuracy and less inference time.

Actions (login required)

View Item