ASLAD-190K: Arabic Sign Language Alphabet Dataset consisting of 190,000 Images

This paper presents ASLAD-190K, a comprehensive dataset comprising 190,000 annotated RGB images representing 32 distinct Arabic Sign Language letters and gestures. Designed to support machine learning and computer vision research, ASLAD-190K offers a diverse range of images collected under varying lighting conditions, distances, and backgrounds to simulate real-world complexities. Random Under-Sampling was applied to address dataset imbalance, yielding a balanced subset of 96,000 images. Additionally, key hand landmarks and 26 geometric angles were extracted to enhance gesture recognition. The dataset is publicly available at https://data.mendeley.com/datasets/2fgpn5dwgc/2 aiming to accelerate the development of assistive technologies for the Arabic-speaking deaf and hard-of-hearing community.