Scaling up sign spotting through sign language dictionaries

TL;DR

This paper introduces a unified learning framework for sign spotting in sign language videos, leveraging weak supervision from subtitles, mouthing cues, and sign language dictionaries, validated on low-shot benchmarks and supported by a new BSL dictionary dataset.

Contribution

It presents a novel integrated approach combining multiple supervision sources for sign spotting and introduces the BSLDict dataset for isolated signs.

Findings

Effective sign spotting on low-shot benchmarks

Successful integration of weak supervision sources

Availability of BSLDict dataset for future research

Abstract

The focus of this work is $\textit{sign spotting}$ - given a video of an isolated sign, our task is to identify $\textit{whether}$ and $\textit{where}$ it has been signed in a continuous, co-articulated sign language video. To achieve this sign spotting task, we train a model using multiple types of available supervision by: (1) $\textit{watching}$ existing footage which is sparsely labelled using mouthing cues; (2) $\textit{reading}$ associated subtitles (readily available translations of the signed content) which provide additional $\textit{weak-supervision}$; (3) $\textit{looking up}$ words (for which no co-articulated labelled examples are available) in visual sign language dictionaries to enable novel sign spotting. These three tasks are integrated into a unified learning framework using the principles of Noise Contrastive Estimation and Multiple Instance Learning. We validate the effectiveness of our approach on low-shot sign spotting benchmarks. In addition, we contribute a machine-readable British Sign Language (BSL) dictionary dataset of isolated signs, BSLDict, to facilitate study of this task. The dataset, models and code are available at our project page.