This repository contains the winning solution developed for the SIH (Smart India Hackathon) Hackathon, focusing on the topic of hand gesture detection using machine learning. The solution utilizes cutting-edge technologies, including Mediapipe, deep learning techniques such as Convolutional Neural Networks (CNNs) and Recurrent Neural Networks (RNNs), and a Google Chrome extension for real-time gesture visualization during virtual meetings.
- Accurate hand gesture detection powered by CNNs and RNNs.
- Seamless integration with Mediapipe for precise gesture analysis.
- Real-time interpretation of both spatial and temporal aspects of hand movements.
- Integration as a Google Chrome extension, enhancing virtual meetings with interactive gestures.
- Transforms conventional video conferencing into engaging and dynamic discussions.
- Mediapipe: Framework for real-time multimedia processing.
- Deep Learning: Utilizes Convolutional Neural Networks (CNNs) and Recurrent Neural Networks (RNNs) for gesture interpretation.
- Google Chrome Extension: Overlaying real-time gesture visualizations onto virtual meetings.
- Clone the repository.
- Install the required dependencies listed in
requirements.txt. - Explore the codebase to understand the implementation of gesture detection and integration with Mediapipe.
- Load the provided pre-trained models for immediate usage or train your own models using the provided datasets.
- Integrate the solution into Google Chrome as an extension following the instructions in the
extension/directory.
Feel free to contribute to the project by opening issues or pull requests. Your feedback and contributions are highly appreciated!