https://github.com/mishig25/3d-posenet

Control 3D Virtual Character through Tensroflow.js Posenet
https://github.com/mishig25/3d-posenet

tensorflow-experiments

Last synced: 2 days ago
JSON representation

Control 3D Virtual Character through Tensroflow.js Posenet

• Host: GitHub
• URL: https://github.com/mishig25/3d-posenet
• Owner: mishig25
• License: mit
• Created: 2018-07-11T17:43:47.000Z (over 6 years ago)
• Default Branch: master
• Last Pushed: 2022-12-03T23:46:26.000Z (about 2 years ago)
• Last Synced: 2024-12-26T06:07:42.278Z (9 days ago)
• Topics: tensorflow-experiments
• Language: JavaScript
• Homepage: https://mishig25.github.io/3d-posenet/
• Size: 20 MB
• Stars: 216
• Watchers: 10
• Forks: 39
• Open Issues: 15
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# 3D-Posenet

### Controlling a 3D Virtual Character Through WebCam

Personal Project

#### Check out the Live Demo: [https://mishig25.github.io/3d-posenet/](https://mishig25.github.io/3d-posenet/)

[`Master Branch`](https://github.com/mishig25/3d-posenet)



[`Experimental Branch`](https://github.com/mishig25/3d-posenet/tree/experimental)



Note: `Master Branch` contains functionalities only for head movements. `Experimental Branch` contains functionalities for arms movements as well; however, the functionalities for arms movements do  not work well enough to be added to `Master Branch` yet.

#### Description:

This web app is an experiment to combine Machine Learning and Computer Graphics through [TensorFlow.js](https://js.tensorflow.org/) and [BabylonJS](https://www.babylonjs.com/). Tensorflow.js is an official Javascript API of Google's popular machine learning framework Tensorflow, and since Tensorflow.js is a Javscript library, it provides a way to run machine learning models in browser environments. On the other hand, BabylonJS is a 3D engine that lets you create and run 3D graphics in web apps. 

### Contents of the repository:

* `app.js` - main React app

* `posenet.js` - class for running Tensorflow.js and Posenet 

* `graphics.js` - class for running BabylonJS and creating the 3D scene

* `joints.js`, `transform.js` - miscellaneous classes

### Development:

```bash

yarn

yarn watch

```

#### License

MIT