Sign Spotting Disambiguation using Large Language Models

TL;DR

This paper presents a training-free, LLM-enhanced framework for sign spotting in sign language videos, improving accuracy and vocabulary flexibility without retraining, by combining feature matching with context-aware disambiguation.

Contribution

The paper introduces a novel, training-free approach that leverages Large Language Models for improved sign spotting and disambiguation in sign language videos, addressing key challenges without retraining.

Findings

Outperforms traditional methods in accuracy and fluency

Demonstrates effectiveness on synthetic and real datasets

Enhances vocabulary flexibility without retraining

Abstract

Sign spotting, the task of identifying and localizing individual signs within continuous sign language video, plays a pivotal role in scaling dataset annotations and addressing the severe data scarcity issue in sign language translation. While automatic sign spotting holds great promise for enabling frame-level supervision at scale, it grapples with challenges such as vocabulary inflexibility and ambiguity inherent in continuous sign streams. Hence, we introduce a novel, training-free framework that integrates Large Language Models (LLMs) to significantly enhance sign spotting quality. Our approach extracts global spatio-temporal and hand shape features, which are then matched against a large-scale sign dictionary using dynamic time warping and cosine similarity. This dictionary-based matching inherently offers superior vocabulary flexibility without requiring model retraining. To mitigate noise and ambiguity from the matching process, an LLM performs context-aware gloss disambiguation via beam search, notably without fine-tuning. Extensive experiments on both synthetic and real-world sign language datasets demonstrate our method's superior accuracy and sentence fluency compared to traditional approaches, highlighting the potential of LLMs in advancing sign spotting.