Speech and written words are examples of natural language used by humans to communicate with one another. Nonetheless, the only way to communicate for the deaf is through sign language. They cannot converge with one another in the absence of an interpreter. For this reason, the development of technology that can read sign language will greatly benefit the social lives of the deaf. Unlike American Sign Language (ASL), Indian Sign Language (ISL) employs both hands for movements rather than just one. Creating an ISL recognition system that translates sign language into legible text is the recommended course of action. There are several ways to accomplish this, including object and position detection, smart gloves, and more. The suggested technique detects hand moments by using TensorFlow object detection. To extract features, MobileNet is utilized in conjunction with single-shot multibox detection (SSD) for further detection.

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

Utilizing Single-Shot Multibox Detection (SSD) in ISL Recognition Systems to Enhance Feature Extraction and Translation Capabilities

  • A Jahnavi,
  • Aparajita Sinha,
  • Monika Agarwal,
  • E. Swetha,
  • Poornima M. Nerale,
  • C. S. S. Krishna Kaushik

摘要

Speech and written words are examples of natural language used by humans to communicate with one another. Nonetheless, the only way to communicate for the deaf is through sign language. They cannot converge with one another in the absence of an interpreter. For this reason, the development of technology that can read sign language will greatly benefit the social lives of the deaf. Unlike American Sign Language (ASL), Indian Sign Language (ISL) employs both hands for movements rather than just one. Creating an ISL recognition system that translates sign language into legible text is the recommended course of action. There are several ways to accomplish this, including object and position detection, smart gloves, and more. The suggested technique detects hand moments by using TensorFlow object detection. To extract features, MobileNet is utilized in conjunction with single-shot multibox detection (SSD) for further detection.