BBC-Oxford British Sign Language Dataset

Samuel Albanie (1) , Gül Varol (2) , Liliane Momeni (3) , Hannah Bull (4, 5) , Triantafyllos Afouras (3) , Himel Chowdhury (3) , Neil Fox (6) , Bencie Woll (6) , Rob Cooper (7) , Andrew Mcparland (7) , Andrew Zisserman (3)

1 CAM - University of Cambridge [UK]
2 LIGM - Laboratoire d'Informatique Gaspard-Monge
3 VGG - Visual Geometry Group
4 LISN - Laboratoire Interdisciplinaire des Sciences du Numérique
5 AMI - Architectures et Modèles pour l'Interaction - LISN
6 DCAL - Deafness, Cognition and Language Research Centre
7 BBC Research & Development

In this work, we introduce the BBC-Oxford British Sign Language (BOBSL) dataset, a large-scale video collection of British Sign Language (BSL). BOBSL is an extended and publicly released dataset based on the BSL-1K dataset introduced in previous work. We describe the motivation for the dataset, together with statistics and available annotations. We conduct experiments to provide baselines for the tasks of sign recognition, sign language alignment, and sign language translation. Finally, we describe several strengths and limitations of the data from the perspectives of machine learning and linguistics, note sources of bias present in the dataset, and discuss potential applications of BOBSL in the context of sign language technology. The dataset is available at https://www.robots.ox.ac.uk/~vgg/data/bobsl/.

Domaines: Vision par ordinateur et reconnaissance de formes [cs.CV]

Samuel Albanie, Gül Varol, Liliane Momeni, Hannah Bull, Triantafyllos Afouras, et al.. BBC-Oxford British Sign Language Dataset. 2022. ⟨hal-03516444⟩