GSL

该数据集是一个实验室制作的集合，包含了10,290个样本，其中包括RGB视频、相应的光泽表示以及希腊书面语言的翻译。此外，该数据集还包含了每个句子的多次重复，没有单一出现或低频词汇，这为模型训练减少了噪音。规模上，10,290个样本对应于331个不同的句子，这些句子由不同的手语者重复表达。该数据集的任务是手语翻译（SLT）。

This dataset is a laboratory-curated collection consisting of 10,290 samples, including RGB videos, corresponding sign language gloss representations, and translations into written Greek. Additionally, each sentence in the dataset has multiple repetitions, with no single-occurrence or low-frequency vocabulary, which helps reduce noise during model training. In terms of scale, the 10,290 samples correspond to 331 distinct sentences, which are repeated by different sign language signers. The task of this dataset is Sign Language Translation (SLT).