Automated sign language digitisation: Leveraging computer vision and archived broadcast footage for accessible avatar generation

Abstract

Sign language serves as the primary linguistic modality for the deaf and hard-of-hearing community. However, the vast majority of sign language content contained within historical video archives remains inaccessible — unindexed, unsearchable and effectively invisible to computational analysis. This paper proposes a novel framework for the automated detection, tracking and semantic annotation of signlanguage motions using advanced computer vision and archived broadcast footage. Building upon initial trials conducted by NHK Enterprises and utilising 2026-era hardware capabilities, the paper evaluates the shift from sensor-based motion capture to videobased pose estimation. Furthermore, it discusses the economic and ethical implications of creating large-scale 3D motion databases, proposing an international consortium model to govern the ‘physical artificial intelligence’ data marketplace. This article is also included in The Business & Management Collection which can be accessed at http://hstalks.com/business.