Ishara-Bochon: The First Multipurpose Open Access Dataset for Bangla Sign Language Isolated Digits
Ishara-Bochon, the first multipurpose comprehensive open access isolated digits dataset for Bangladeshi Sign Language (BdSL) is introduced in this article. Sign language digits is a major part of communication for deaf and hearing impaired people. It is important in their everyday stuff, such as – in routine accounting and to interact with general people. Ishara-Bachon dataset contains 100 sets of 10 Bangla basic sign digits (0, 1, 2 . . . 9), collected from different deaf and general volunteers from different institutes. After discarding maximum errors and performing different preprocessing methods, 1000 images of Bangla sign language isolated digits were included in the first dataset. After collectiong raw data, some effective preprocessing methods were performed for making those data useable for computer vision model and any other application development purposes.
Bangla Sign Language, NLP, Computer vision, Machine Learning, Image processing, Sign language digits, BdSL, Open source data, Pattern recognition
Md. Sanzidul Islam, Sadia Sultana Sharmin Mousumi, Nazmul A. Jessan, AKM Shahariar Azad Rabby, Sheikh Abujar, Syed Akhter Hossain