ISLAN

Published: 8 January 2021| Version 1 | DOI: 10.17632/rc349j45m5.1
Contributors:
Elakkiya R,

Description

In order to help the deaf-mute community, in this research work we have collected and created a dataset for Indian Sign Language (ISL) Recognition called the ISLAN dataset. ISLAN consists of the ISL representation of English Alphanumeric signs. ISLAN has a total collection of 700 sign images and 24 sign videos. ISLAN dataset is intended to be publicly made available for free for the research community so as to help the impaired community to enable the communication by developing the ISL recognition system.

Files

Steps to reproduce

The mute communities are totally dependent on sign language to communicate with others while the rest of the population is not able to understand and interpret it. The sign language recognition system tends to bridge this communication gap between them and the rest of the community. Currently, there is a lack of publicly available datasets for Indian signs to develop ISLR system. We introduce an image dataset called ISLAN, for ISLR which comprises of the Indian sign language representation of English Alphabets and Numbers. ISLAN is a collection of a total of 700 sign images and 24 videos. Each image and video illustrates the sign language translation of either an English alphabet or a number. For alphabets, we have created and collected the dataset based on both single handed and double handed signer representations. All the images are in JPEG format. Few of the Alphabetical representations were action gestures (J, Z) hence we have collected videos for those alphabet signs and the rest of the alphabets had static gestures for which images have been captured. ISLAN has 350 unique images and 12 unique videos captured with the help of 6 volunteer signers. This dataset is organized into separate collections based on the signers, letters, and single-handed or double-handed gestures. To differentiate between the single-handed and double-handed signs for alphabets we have organized them separately. This data can be used by any researchers to recognise Indian Signs and various computer vision and deep learning techniques can be used to further proceed with better recognition results.

Institutions

Shanmugha Arts Science Technology and Research Academy School of Computing

Categories

Computer Vision, Sign Language, Deep Learning

Licence