Amharic Sign Language Recognition

This project is an Amharic sign language recognition system that uses a webcam to detect hand gestures and predict corresponding Amharic letters in real-time. The project leverages Mediapipe for hand landmark detection, OpenCV for capturing video frames and drawing bounding boxes, and a Random Forest Classifier trained on hand landmarks for letter classification.

Features

• Real-time hand gesture recognition using a webcam.
• Prediction of Amharic letters based on hand landmarks.
• Custom Amharic font rendering using the PIL library to display predictions on the screen.

Prerequisites

Before you begin, ensure you have the following installed on your system:

• Python 3.x
• Mediapipe (mediapipe library)
• OpenCV (opencv-python)
• PIL (Pillow)
• NumPy
• Scikit-learn
• A trained classifier model stored in a .p file (Pickle format)

You can install the required Python packages with:

pip install mediapipe opencv-python pillow numpy scikit-learn

installation

• git clone https://github.com/Naoldaba/Amharic-SignLanguageRecognition.git
• cd Amharic-SignLanguageRecognition

Running the Model

• python sign_language_predictor.py