How2Sign: A Large-scale Multimodal Dataset for Continuous American Sign Language.

ID: 4108
School: School of Language, Education, and Culture
Program: American Sign Language
Status: Completed
Start date: January 2021
End Date: December 2021

Description

One of the factors that have hindered progress in the areas of sign language recognition, translation, and production is the absence of large annotated datasets. Towards this end, we introduce How2Sign, a multimodal and multiview continuous American Sign Language (ASL) dataset, consisting of a parallel corpus of more than 80 hours of sign language videos and a set of corresponding modalities including speech, English transcripts, and depth. A three-hour subset was further recorded in the Panoptic studio enabling detailed 3D pose estimation. To evaluate the potential of How2Sign for real-world impact, we conduct a study with ASL signers and show that synthesized videos using our dataset can indeed be understood. The study further gives insights on challenges that computer vision should address in order to make progress in this field.

Principal investigators

Priorities addressed

Approved Products

2021

De Haan, K., Duarte, A., Ghadiyaram, D., Giro-i-Nieto, X., Metze, F., Palaskar, S., Torres, J., & Ventura, L. (2021, June 17). How2Sign: A Large-scale Multimodal Dataset for Continuous American Sign Language. In the 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), June 17, 2021, Virtually. http://cvpr2021.thecvf.com