A Dataset for recognition of Norwegian Sign Language

Abstract

Machine learning is a powerful tool in developing sign language recognition models, significantly enhancing communication accessibility for the deaf and hard-of-hearing community. However, such models' creation relies on the availability of comprehensive datasets, which are currently scarce for specific sign languages, particularly for Norwegian Sign Language (NSL). This paper introduces a unique dataset created to address this gap. The dataset, comprising 24,300 images of 27 NSL letters, was captured under varying conditions to represent each sign comprehensively. A man and a woman performed the signs. The goal of creating this dataset was to provide a robust foundation for further research in NSL recognition.

The dataset is hosted by the Department of Applied Data Science at Noroff University College, Norway, and is freely accessible [1] at https://data.mendeley.com/datasets/3cfrj4vd4m/1.