YouTube-ASL: A Large-Scale, Open-Domain American Sign Language-English Parallel Corpus

David Uthus, Garrett Tanzer, Manfred Georg

Keywords: Sign language translation, dataset

TL;DR: Introducing a new datasets composed of YouTube videos for American Sign Language to English translation.

Abstract: Machine learning for sign languages is bottlenecked by data. In this paper, we present YouTube-ASL, a large-scale, open-domain corpus of American Sign Language (ASL) videos and accompanying English captions drawn from YouTube. With ~1000 hours of videos and >2500 unique signers, YouTube-ASL is ~3x as large and has ~10x as many unique signers as the largest prior ASL dataset. We train baseline models for ASL to English translation on YouTube-ASL and evaluate them on How2Sign, where we achieve a new fine-tuned state of the art of 12.397 BLEU and, for the first time, nontrivial zero-shot results.

Submission Number: 951