Fastai → TensorFlow.js → Unity Tutorial
Tutorial Links
- Part 1: Train a hand gesture classifier using fastai and export it to TensorFlow.js.
- Part 2: Create a TensorFlow.js plugin for the Unity game engine.
- Part 3: Build a Unity project as a WebGL application and host it using GitHub Pages.
In-Browser Demo
| GitHub Repository | Demo Link |
|---|---|
| fastai-hand-gesture-classifier-webgl-demo | Hand Gesture Classifier |
Training Code
| Jupyter Notebook | Colab | Kaggle |
|---|---|---|
| GitHub Repository |  |
 |
Kaggle Datasets
- HaGRID Classification 512p no_gesture 150k
- Images: 154,816
- Size: 4 GB
- HaGRID Classification 512p no_gesture 300k
- Images: 309,632
- Size: 8 GB
- HaGRID Classification 512p no_gesture
- Images: 619,264
- Size: 15 GB
Reference Images
| Class | Image |
|---|---|
| call |  |
| dislike |  |
| fist |  |
| four |  |
| like |  |
| mute |  |
| ok |  |
| one |  |
| palm |  |
| peace |  |
| peace_inverted |  |
| rock |  |
| stop |  |
| stop_inverted |  |
| three |  |
| three2 |  |
| two_up |  |
| two_up_inverted |  |