The ISW-1000 dataset
ISW-1000 (1000 Isolated Sign Language by Leshan Normal University) is a dataset captured in a laboratory environment, currently consisting of 1000 signs, each captured by ten signers, resulting in a total of 10,000 videos. All signers were instructed to stand and the background was green. Since our goal is to build a standardized dataset, we did not consider the deviations caused by subjective factors of signers in real applications during the shooting process. Written informed consent was obtained from all the participants prior to the enrollment (or for the publication) of this study.
The signers include 8 deaf students and 2 hard-of-hearing students, all proficient in the national general sign language of China. These students have received systematic sign language education and focused on the standardization of sign language during the shooting, avoiding personal styles. Additionally, there was a professional sign language teacher present during the shooting to provide guidance.
All videos were uniformly shot with only the upper body of the signer and a green background, ensuring that no objects obstruct the signer’s hands. We considered using professional cameras or smartphones for capturing, but considering that our shooters are students, using smartphones seemed more relevant and convenient for future dataset expansion.
The selected sign words are drawn from the national universal sign language dictionary of China, specifically those that appear most frequently on Wikipedia. We will provide label files for each video.
The dataset is released under the CC-BY license (
CC BY 4.0).
All technical papers, documents and reports which use the ISW-1000 database will acknowledge the use of the database by giving a citation to “Yazhou Ren, Hongkai Li, Yuhao Li, Jingyu Pu, Xiaorong Pu, Siyuan Jing, Peng Jin, Lifang He. Multi-modal Isolated Sign Language Recognition Based on Self-paced Learning. Expert Systems with Applications, 2025, 291: 128340”. Tha paper can be downloaded from: https://authors.elsevier.com/a/1lFNZ3PiGTXIB7
@article{ren2025multi,
title={Multi-modal Isolated Sign Language Recognition Based on Self-paced Learning},
author={Ren, Yazhou and Li, Hongkai and Li, Yuhao and Pu, Jingyu and Pu, Xiaorong and Jing, Siyuan and Peng, Jin and He, Lifang},
journal={Expert Systems with Applications},
volume = {291},
pages={128340},
year={2025},
publisher={Elsevier}
}
download ISW-1000.zip
正文