Labeeb

Labeeb is the first Lebanese Sign Language translation app, developed to help bridge communication between sign language users and people who do not understand Lebanese Sign Language. The app provides real-time offline translation for the Arabic alphabet and common Lebanese signs.

I built the system end-to-end, including data collection, machine learning models, backend API, and mobile application. The recognition pipeline uses MediaPipe landmarks with lightweight TensorFlow models, enabling efficient sign recognition without relying directly on raw image data.

The application combines:

• Real-time offline sign recognition, allowing translation without requiring a continuous internet connection.
• Landmark-based recognition using MediaPipe and TensorFlow.
• Static and dynamic sign classification for different types of signs.
• Flask REST API for backend functionality and communication with the application.
• Flutter mobile frontend for Android and iOS.
• Custom sign-language data collection and preprocessing used to develop the recognition models.

The work on Labeeb also led to a research preprint on lightweight Arabic Sign Language recognition using landmark-based representations, exploring efficient models for both static and dynamic sign recognition.

Project link: https://zeinshehab.github.io/Labeeb/

• Machine Learning
• TensorFlow
• MediaPipe
• Computer Vision
• Python
• Flask
• REST APIs
• Flutter
• Android
• iOS
• Mobile Development