This project aims to classify sign language letters using machine learning. This README provides an overview of the project and instructions for usage.
The project consists of five main files:
augmentation.py: This file contains code for data augmentation, which is a technique used to increase the size and diversity of a dataset.
collect_images.py: This file uses OpenCV to collect images of sign language letters from the user's camera.
create_data.ipynb: This Jupyter Notebook uses MediaPipe to track the hand in the collected images and extract the keypoint features. The keypoint features are then used to create a dataset for training the random forest classifier.
train_model.py: This file trains the random forest classifier on the created dataset.
inference.py: This file is used to classify sign language letters in real-time.
You only need two files to run the project; utils/model.pkl and inference.py. Here are the main usage instructions:
-
Run the real-time sign langugae detection script:
python inference.py -
A live video stream will open, and the application will start detecting hands and their status.
-
To quit the application, press the 'q' key.