Sign-Language Datasets at Scale: A Comprehensive Survey on Resources, Benchmarks, and Annotation Standards
A new survey paper published on arXiv details the current landscape of sign language datasets, highlighting significant challenges in their development and utilization. The paper identifies fragmented datasets, inconsistent annotations, and limited linguistic coverage as major hurdles in advancing sign language recognition and translation technologies. To address these issues, the authors have compiled a comprehensive index of 120 resources across 35 sign languages and introduced a standardized 24-field Sign-Language Datasheet, along with a public GitHub repository to promote reproducible evaluation and inclusive technology development. AI
IMPACT This survey aims to standardize documentation and evaluation for sign language technologies, potentially accelerating the development of more inclusive and robust AI systems for the deaf and hard-of-hearing communities.