This project demonstrates real-time object detection using a webcam, powered by TensorFlow.js and the COCO-SSD (Common Objects in Context - Single Shot MultiBox Detector) pretrained model. It allows users to see live predictions of objects in the camera feed directly through their web browser.
- HTML: Provides the structure of the webpage.
- CSS: Adds basic styling to the webpage elements.
- JavaScript (TensorFlow.js): Integrates the COCO-SSD model and manages the webcam feed.
- TensorFlow.js (COCO-SSD): Pretrained model for object detection.
No installation is required other than a modern web browser with webcam support.
- Clone the repository or download the project files.
- Open
index.html
in a web browser that supports WebGL and WebRTC (e.g., Google Chrome). - Allow access to your webcam when prompted.
- The webcam feed will start, and objects detected by COCO-SSD will be outlined in real-time.
This project can be implemented in various real-life scenarios:
- Security: Monitoring for specific objects or intrusions.
- Retail: Tracking customer movements or inventory management.
- Education: Interactive demonstrations of object recognition.
A live demo of this project can be found here (add the link to your live demo if available).
Contributions are welcome! Feel free to fork the repository and submit pull requests.
This project is licensed under the MIT License.
- TensorFlow.js and the COCO-SSD model for providing powerful tools for machine learning in the browser.
- Inspiration and initial code structure from TensorFlow.js examples.
Feel free to customize this README file further based on specific details of your project and implementation. Adjust the sections like Demo, Acknowledgments, and Installation based on your project's actual details and preferences.