ISLTranslate

ISLTranslate是由印度理工学院坎普尔分校和海得拉巴分校共同创建的一个大型数据集，专注于连续印度手语（ISL）到英语的翻译。该数据集包含31,222对ISL-英语句子和短语，覆盖了日常交流中的广泛词汇，共有11,655个词汇。数据集主要来源于ISLRTC和Deaf Enabled Foundations提供的教育视频，旨在为听力障碍儿童提供基础教育。创建过程中，使用先进的语音到文本模型自动生成文本，并进行人工校验。ISLTranslate的应用领域主要集中在改善听力障碍社区与主流社会之间的沟通，通过开发高效的统计手语翻译系统来解决沟通障碍问题。

ISLTranslate is a large-scale dataset co-developed by the Indian Institute of Technology Kanpur and the Indian Institute of Technology Hyderabad, focusing on continuous Indian Sign Language (ISL) to English translation. This dataset contains 31,222 pairs of ISL-English sentences and phrases, covering a wide range of daily communication vocabulary with a total of 11,655 distinct lexical items. It is primarily sourced from educational videos provided by ISLRTC and Deaf Enabled Foundations, and was created with the goal of providing basic education for children with hearing impairments. During its development, advanced speech-to-text models were used to automatically generate text transcripts, followed by manual verification and correction. The main application scenarios of ISLTranslate center on improving communication between the hearing-impaired community and the mainstream society, addressing communication barriers by developing efficient statistical sign language translation systems.