Active Learning for Multilingual Fingerspelling Corpora

TL;DR

This paper explores the use of active learning and pre-training to address data scarcity in multilingual fingerspelling sign language corpora, analyzing the impact of visual and linguistic similarities across languages.

Contribution

It introduces a novel analysis of pre-training effects in multilingual sign language recognition, highlighting the role of visual versus linguistic similarities.

Findings

Pre-training benefits are observed in fingerspelling recognition.

Visual similarities may influence pre-training effectiveness more than linguistic ones.

Active learning helps mitigate data scarcity in sign language corpora.

Abstract

We apply active learning to help with data scarcity problems in sign languages. In particular, we perform a novel analysis of the effect of pre-training. Since many sign languages are linguistic descendants of French sign language, they share hand configurations, which pre-training can hopefully exploit. We test this hypothesis on American, Chinese, German, and Irish fingerspelling corpora. We do observe a benefit from pre-training, but this may be due to visual rather than linguistic similarities