Transferring Cross-domain Knowledge for Video Sign Language Recognition

TL;DR

This paper introduces a novel approach for video sign language recognition that leverages subtitled news videos by transferring domain-invariant visual concepts, significantly improving recognition accuracy despite domain gaps.

Contribution

The paper proposes a method to transfer knowledge from subtitled sign news videos to improve word-level sign language recognition models, addressing data annotation challenges.

Findings

Outperforms previous state-of-the-art methods on standard datasets.

Achieves 28.1 AP@0.5 in sign localization.

Effectively transfers knowledge across domain gaps.

Abstract

Word-level sign language recognition (WSLR) is a fundamental task in sign language interpretation. It requires models to recognize isolated sign words from videos. However, annotating WSLR data needs expert knowledge, thus limiting WSLR dataset acquisition. On the contrary, there are abundant subtitled sign news videos on the internet. Since these videos have no word-level annotation and exhibit a large domain gap from isolated signs, they cannot be directly used for training WSLR models. We observe that despite the existence of a large domain gap, isolated and news signs share the same visual concepts, such as hand gestures and body movements. Motivated by this observation, we propose a novel method that learns domain-invariant visual concepts and fertilizes WSLR models by transferring knowledge of subtitled news sign to them. To this end, we extract news signs using a base WSLR model, and then design a classifier jointly trained on news and isolated signs to coarsely align these two domain features. In order to learn domain-invariant features within each class and suppress domain-specific features, our method further resorts to an external memory to store the class centroids of the aligned news signs. We then design a temporal attention based on the learnt descriptor to improve recognition performance. Experimental results on standard WSLR datasets show that our method outperforms previous state-of-the-art methods significantly. We also demonstrate the effectiveness of our method on automatically localizing signs from sign news, achieving 28.1 for AP@0.5.