ASLLVD (American Sign Language Lexicon Video Dataset)

美国手语词典视频数据集 (ASLLVD) 由引用形式的> 3,300个ASL符号的视频组成，每个由1-6个本地ASL签名者制作，总共近9,800个令牌。该数据集包括多个同步视频，从不同角度显示签名。语言注释包括光泽标签，符号开始和结束时间代码，双手的开始和结束手形标签，符号类型的形态和发音分类。对于复合符号，数据集包括每个语素的注释。为了促进基于计算机视觉的手语识别，数据集还包括用于符号变体的数字ID标签，未压缩原始格式的视频序列以及相机校准序列。

American Sign Language Lexicon Video Dataset (ASLLVD) comprises videos of over 3,300 ASL signs, with each sign's footage produced by 1 to 6 native ASL signers, totaling nearly 9,800 tokens. The dataset includes multiple synchronized videos capturing the sign from different perspectives. Linguistic annotations cover gloss labels, start and end time codes of the sign, start and end handshape labels for both hands, as well as morphological and articulatory classifications of sign types. For compound signs, the dataset provides annotations for each individual morpheme. To facilitate computer vision-based sign language recognition, the dataset also offers numeric ID labels for sign variants, video sequences in uncompressed raw format, and camera calibration sequences.

搜集汇总

背景与挑战

背景概述

ASLLVD是一个包含3,300个美国手语符号视频的数据集，总计约9,800个令牌，提供多角度视频和详细的语言注释，适用于手语识别研究。数据集由波士顿大学于2012年发布。