Indian Sign Language Image Dataset
This repository contains a 12,700 strong image dataset containing ISL signage of the 26 English alphabets. There are 26 classes and 486 processed and augmented images per class and these images can undergo further augmentation to boost the size and increase the images per class. Further classes can also be added (i.e numbers, words, etc.) to increase the research scope.
Also added is the train and test csv files of the dataset split into 20-80 as of December 6, 2020. These csv files have been shuffled and labeled with each row representing the pixel values of each image. The dataset has been successfully used for Image Recognition project using Neural Networks with great accuracy metrics.
Big thanks to LEELA SURYA TEJA MANGAMURI, friends and family for contribution of images to the dataset. As well as fellow team members Esha Gavali, Shivani Kamtikar and Urvi Lendhe for working on this project. A lot of this dataset was self-generated to as no such ISL dataset was available at the time of inception of our project.