---

Bangla Sign Language Alphabet Recognition Using Hand Gestures : A Deep Learning Approach

Corresponding Author : T. B. Das (taanmoycse15@gmail.com)

Authors : T. B. (taanmoycse15@gmail.com)

Keywords : Bangla Sign Language (BSL), Convolutional Neural Network (CNN), NASnet, Segmentation, Feature extraction

Abstract :

Real-time recognition of Bangla Sign Language (BSL) alphabet using hand gestures is a fascinating research topic for image processing and computer vision enthusiasts. A significant amount of researches has been done on this topic recently. BSL is a blessing for deaf and mute people of the Bengali community around the world as it minimizes discomfort in their day-to-day lives. However, in most cases, people without disabilities do not know Bangla Sign Language, thus there is a huge barrier to communication between the hearing-impaired people and normal people. The purpose of this research is to provide insights into various methods and propose an improved approach for recognizing Alphabets using hand gestures in BSL. Our proposed recognition system using Convolutional Neural Network (CNN) provides an accuracy of 97.2% for real-time prediction of BSL alphabets using hand gestures. The BSL alphabet recognizer can be used in practical life as a translator for communication between hearing-impaired people and normal people.

Published on March 7th, 2021 in Volume 2 Issue 1 & 2, Computer Science, Electrical and Electronics