This is a dynamic sign language detector. By using TensorFlow, MediaPipe and OpenCV, we're able to train a model to accurately recognize dynamic sign language gestures such as Hello, I Love You and Thanks.
- Real-time detection of dynamic sign gestures
- Trained using a custom dataset with keypoint extraction
- Uses webcam input for live gesture recognition
- High accuracy
-
Python
-
TensorFlow/Keras for gesture classification model
-
OpenCV for video capture and visualization
-
MediaPipe for hand, pose, and face landmark detection
-
NumPy for array manipulation
Project_SilentTalks/
├── Elites Final/ # Contains initial files for website
├── MP_data/ # Contains local training data used for initial model training
├── action.h5 # Trained Keras model for gesture classification
├── Action_Detection_Utils.py # Utility functions: drawing landmarks, extracting keypoints, etc.
├── Action Detection Refined.ipynb # Model training notebook using collected data
├── README.md # Project overview and instructions
├── Runner.py # Script for running the project
├── Demo.gif # GIF of the project in action
└── requirements.txt # Required Python libraries
- Clone the repository and navigate to it on your machine
git clone https://github.com/Im-Arth1307/Project_SilentTalks
cd Project_SilentTalks- Install all the necessary dependencies
pip install -r requirements.txt
- Run the project
python Runner.py
- Press 'Q' on your keyboard to close the sign language detection window
Real-time recognition of "Hello", "ThankYou" and "ILoveYou"
- Add more gestures and words
- Expand the dataset by extracting keypoint data from publicly available sign language videos
- Finish integration and deployment as a web app
Contributions are always welcome ! Feel free to fork the repo, make improvements and create a pull request.
For any questions or suggestions, reach out to:
Atharva Jakhetiya
Gmail: [email protected]