Continuous Sign Language Recognition System using Deep Learning with MediaPipe Holistic

  • 2024-11-07 08:19:39
  • Sharvani Srivastava, Sudhakar Singh, Pooja, Shiv Prakash
  • 0

Abstract

Sign languages are the language of hearing-impaired people who use visualslike the hand, facial, and body movements for communication. There aredifferent signs and gestures representing alphabets, words, and phrases.Nowadays approximately 300 sign languages are being practiced worldwide such asAmerican Sign Language (ASL), Chinese Sign Language (CSL), Indian Sign Language(ISL), and many more. Sign languages are dependent on the vocal language of aplace. Unlike vocal or spoken languages, there are no helping words in signlanguage like is, am, are, was, were, will, be, etc. As only a limitedpopulation is well-versed in sign language, this lack of familiarity of signlanguage hinders hearing-impaired people from communicating freely and easilywith everyone. This issue can be addressed by a sign language recognition (SLR)system which has the capability to translate the sign language into vocallanguage. In this paper, a continuous SLR system is proposed using a deeplearning model employing Long Short-Term Memory (LSTM), trained and tested onan ISL primary dataset. This dataset is created using MediaPipe Holisticpipeline for tracking face, hand, and body movements and collecting landmarks.The system recognizes the signs and gestures in real-time with 88.23% accuracy.

 

Quick Read (beta)

loading the full paper ...