Visit the working prototype here
To imitate a system that could in real-time convert sign language to voice. Such a system would be of immense use for the mute and deaf community when it comes to interactions in their daily life.
A simple platform that takes in a photo and predicts the Alphabet represented in the photo taken.
- Not real-time
- Azure Based Limitations (5000 images training only)
- Current AI Model not that accurate
- Azure Custom Vision (Prediction and Training)
- Azure Web App Service for Hosting
The ASL dataset from Kaggle was used for this project. Not all the images were used but 190 images of all the 26 alphabets were taken randomly and used for training the azure custom vision model.