This web game leverages deep learning models for pose estimation and action recognition, delivering real-time feedback based on accurately detected poses.
The application integrates the following functionalities:
- Pose Estimation: Utilizes the PoseNet model to estimate human poses in real-time.
- Action Recognition: Employs a pre-trained Teachable Machine classification model to recognize specific actions based on the detected poses.
- Interactive Feedback: Provides interactive feedback by triggering animations and updating scores and actions on the user interface.
Play it online or clone the project to run it locally through a live server on a web browser. Once the application is running, access it through a web browser. The webcam feed will display in real-time, and the application will recognize and respond to different poses and interactions with the 04 colored circles on the screen.
- HTML, CSS and Javascript
- TensorFlow.js: Utilized for loading and running the pre-trained machine learning models.
- Teachable Machine: Provides a pre-trained classification model for action recognition.
- PoseNet: Enables real-time human pose estimation from input images or video.
Contributions to the project are welcome. To contribute, follow these steps:
- Fork the repository.
- Create a new branch for your feature or bug fix:
git checkout -b feature-name - Make changes and commit them:
git commit -m 'Description of changes' - Push to the branch:
git push origin feature-name - Submit a pull request.
This project is licensed under the MIT License.
This project was developed by Luciano Ayres.
- Special thanks to the developers of TensorFlow.js, Teachable Machine, and PoseNet for their valuable contributions to the machine learning community.