Dataset Details
Title:
SignTalk-GSL: Ghanaian Medical Sign Language Data
Details:
A large-scale, categorized dataset of 10,000 sign language videos for healthcare
SignTalk-GH is a dataset of 10,000 Ghanaian Sign Language (GhSL) videos tailored to healthcare settings. The foundation of this dataset lies in a carefully curated pool of 4,000+ unique sentences designed to capture the breadth of clinical interactions in Ghana, spanning general consultations, pediatrics, pharmacy visits, mental health support, dermatology, and more.
To ensure semantic diversity and cultural relevance, these sentences were organized into 26 thematic categories, with attention to the available GhSL vocabulary. The sentences were collaboratively developed by a multidisciplinary team of healthcare professionals, certified GhSL interpreters, and linguists.
Each sentence was signed and recorded by multiple signers, resulting in 10,000 video samples. This makes SignTalk-GH an ideal resource for building AI models for sign language recognition, translation, and assistive healthcare technologies, and for linguistic, cultural, and accessibility research in West African healthcare communication.
File name
Description
Videos/
Directory containing 10,000 video files of signed sentences
Metadata.csv
Metadata file detailing sentence Id, sentence text, and sentence category.
README.md
Documentation, dataset description, and usage guidelines