BosphorusSign22k Sign Language Recognition Dataset

BosphorusSign22k is a benchmark dataset for vision-based, user-independent sign language recognition (Sign Language Recognition). This dataset is based on the BosphorusSign dataset and is designed to help the linguistics and computer science communities.

This dataset contains Turkish Sign Language videos from three different domains: health, finance, and common everyday signs. The videos in this dataset are performed by six native signers, facilitating user-independent research on sign language.