🤖 AI Summary
Visible-light video-based sign language translation suffers from sensitivity to illumination variations, motion blur, and privacy leakage. To address these limitations, this work pioneers the integration of event cameras into sign language translation, introducing Event-CSL—the first large-scale, high-resolution Chinese event-stream sign language benchmark comprising 14,827 samples, captured under diverse viewpoints, lighting conditions, and dynamic gesture scenarios. We propose a CNN-Mamba collaborative architecture: CNNs efficiently extract spatially sparse features from asynchronous event streams, while the Mamba module captures long-range temporal dependencies. Experiments demonstrate substantial improvements in translation robustness and accuracy, achieving a +12.6 BLEU gain over video-based baselines on Event-CSL. Both the code and dataset are fully open-sourced, establishing foundational resources to advance standardization of event-driven sign language translation.
📝 Abstract
Sign Language Translation (SLT) is a core task in the field of AI-assisted disability. Unlike traditional SLT based on visible light videos, which is easily affected by factors such as lighting, rapid hand movements, and privacy breaches, this paper proposes the use of high-definition Event streams for SLT, effectively mitigating the aforementioned issues. This is primarily because Event streams have a high dynamic range and dense temporal signals, which can withstand low illumination and motion blur well. Additionally, due to their sparsity in space, they effectively protect the privacy of the target person. More specifically, we propose a new high-resolution Event stream sign language dataset, termed Event-CSL, which effectively fills the data gap in this area of research. It contains 14,827 videos, 14,821 glosses, and 2,544 Chinese words in the text vocabulary. These samples are collected in a variety of indoor and outdoor scenes, encompassing multiple angles, light intensities, and camera movements. We have benchmarked existing mainstream SLT works to enable fair comparison for future efforts. Based on this dataset and several other large-scale datasets, we propose a novel baseline method that fully leverages the Mamba model's ability to integrate temporal information of CNN features, resulting in improved sign language translation outcomes. Both the benchmark dataset and source code will be released on https://github.com/Event-AHU/OpenESL