SignLens - Deep Learning for Sign Language Classification

SignLens is a project that leverages the power of Deep Learning to classify American Sign Language (ASL) gestures. The project includes scripts for data preprocessing, model training, and prediction. It is a final project for the Data Science bootcamp at Le Wagon.

This repository hosts the frontend for the intuitive Streamlit interface, providing seamless accessibility: SignLens Streamlit Interface

Here is a preview:

SignLens.mp4

Deep Learning Classification Model

SignLens uses an approach where videos are initially transformed into JSON files containing Mediapipe landmarks. Subsequently, these JSON files are transmitted to an API housing the landmark model, which returns accurate classifications. The development of both the model and the API is detailed here.

Authors

• Benoît Frisque
• Wail Benrabh
• Jan Storz
• Maximilien Grieb

Contact

If you have any questions, comments, or feedback, feel free to reach out to us!

• Email: [email protected]
• GitHub: benoitfrisque