American Sign Language Translator
Problem Statement
- The people having speaking difficulties or hearing disability have a hard time in expressing thoughts and communicating with other people .
- As a result the social, emotional and linguistic growth of such people is hampered and they develop anxiety and depression symptoms.
Solution
- Integrating hand gesture recognition system using computer vision for establishing 2-way communication system.
- To provide information access and services to deaf and dumb people in American sign language.
- The project is scalable and has been extended to capture whole vocabulary of ASL through manual and non manual signs.
Methodology
- Hand Detection using YOLO, CNN and OpenCV.
- FFmpeg to extract hand detected from the video and then Deep CNN was used for mapping the hand gestures recognized to the nearest alphabet in American sign language(ASL).
- Converted the predicted text to speech to make it more interpretable.
IMPLEMENTATION
(ALL ALPHABETS)
1.00
IMPLEMENTATION
(TEST WORD)
Web Implementation
(Using NodeJS for backend and SocketIO for real-time communication)
1.00
Future Goals
-
Real time translation from English to other languages as well.
- Use Facial Detection to extract the user's expression from the video and use it to change tone of predicted speech.