Bornil:一个用于 AI 赋能的方言无关交流的开源手语数据众包平台
人机交互
2023-08-30 v1
摘要
标注手语数据集的缺失阻碍了手语识别与翻译技术的发展。本文介绍Bornil;一个对众包友好、多语言的手语数据采集、标注与验证平台。Bornil允许用户录制手语手势,并让标注者进行句子级与词 gloss 级标注。它还允许验证者通过人工验证确保所录视频与标注的质量,从而为基于深度学习的自动手语识别构建高质量数据集。为展示系统效能,我们采集了规模最大的孟加拉手语方言数据集,进行了基于深度学习的手语识别建模,并报告了基准性能。Bornil平台、BornilDB v1.0数据集及代码库可在 https://bornil.bengali.ai 获取。
引用
@article{arxiv.2308.15402,
title = {Bornil: An open-source sign language data crowdsourcing platform for AI enabled dialect-agnostic communication},
author = {Shahriar Elahi Dhruvo and Mohammad Akhlaqur Rahman and Manash Kumar Mandal and Md. Istiak Hossain Shihab and A. A. Noman Ansary and Kaneez Fatema Shithi and Sanjida Khanom and Rabeya Akter and Safaeid Hossain Arib and M. N. Ansary and Sazia Mehnaz and Rezwana Sultana and Sejuti Rahman and Sayma Sultana Chowdhury and Sabbir Ahmed Chowdhury and Farig Sadeque and Asif Sushmit},
journal= {arXiv preprint arXiv:2308.15402},
year = {2023}
}
备注
6 pages, 7 figures