Open AccessSBC Based Object and Text Recognition Wearable System u sing Convolutional Neural Network with Deep Learning Algorithm
Author(s) -
Melchiezhedhieck J. Bongao,
Arvin F. Almadin,
Christian L. Falla,
Juan Carlo F. Greganda,
Steven Valentino E. Arellano,
Phillip Amir M. Esguerra
Publication year - 2021
Publication title -
international journal of recent technology and engineering
Language(s) - English
Resource type - Journals
ISSN - 2277-3878
DOI - 10.35940/ijrte.c6474.0910321
Subject(s) - python (programming language) , computer science , arduino , convolutional neural network , wearable computer , artificial intelligence , waterfall model , classifier (uml) , deep learning , object (grammar) , computer vision , speech recognition , machine learning , software , programming language , embedded system
This Raspberry Single-Board Computer-Based Object and Text Real-time Recognition Wearable Device using Convolutional Neural Network through TensorFlow Deep Learning, Python and C++ programming languages, and SQLite database application, which detect stationary objects, road signs and Philippine (PHP) money bills, and recognized texts through camera and translate it to audible outputs such as English and Filipino languages. Moreover, the system has a battery notification status using an Arduino microcontroller unit. It also has a switch for object detection mode, text recognition mode, and battery status report mode. This could fulfill the incapability of visually impaired in identifying of objects and the lack of reading ability as well as reducing the assistance that visually impaired needs. Descriptive quantitative research, Waterfall System Development Life Cycle and Evolutionary Prototyping Models were used as the methodologies of this study. Visually impaired persons and the Persons with Disability Affairs Office of the City Government of Biñan, Laguna, Philippines served as the main respondents of the survey conducted. Obtained results stipulated that the object detection, text recognition, and its attributes were accurate and reliable, which gives a significant distinction from the current system to detect objects and recognize printed texts for the visually impaired people.