Scholarlink

Sign language is the easiest way to communicate between deaf people. There have been many studies on sign language recognition by computer vision, which, however, is not privacy-friendly and lacks user-friendliness. So, how can deaf individuals actively convey their meaning to people who do not understand sign language in an easy and practical way? In this paper, we propose a new method that can achieve this. The method involves a wearable device equipped with flex sensors and an accelerometer, combined with an LSTM model to recognize the words of people who are unable to speak. The model was trained and tested with 15 sign languages, including static and dynamic gestures. Finally, this project has been successfully implemented with 98% accuracy based on a training dataset recorded by the author within one hour and is capable of translating 15 word-level sign languages into speech in real time.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Smart Glove for Word-Level Sign Language Recognition Using Flex Sensors, Accelerometer Sensors, and Long-Short Term Memory Model

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Minh Chanh Vo,
  • Trong Nhan Le

摘要

Sign language is the easiest way to communicate between deaf people. There have been many studies on sign language recognition by computer vision, which, however, is not privacy-friendly and lacks user-friendliness. So, how can deaf individuals actively convey their meaning to people who do not understand sign language in an easy and practical way? In this paper, we propose a new method that can achieve this. The method involves a wearable device equipped with flex sensors and an accelerometer, combined with an LSTM model to recognize the words of people who are unable to speak. The model was trained and tested with 15 sign languages, including static and dynamic gestures. Finally, this project has been successfully implemented with 98% accuracy based on a training dataset recorded by the author within one hour and is capable of translating 15 word-level sign languages into speech in real time.