#338 · Primary category: Computer Vision

QuickDraw

cnn computer-vision cv2 deep-learning deep-neural-networks deeplearning image-classification neural-network neural-networks opencv opencv-python python python3 pytorch quickdraw quickdraw-dataset

Implementation of Quickdraw - an online game developed by Google

Project last updated:01/11/23

GitHub Stars

1.1K

Forks

195

Contributors

1

License

MIT

Why we included this project

Sketch recognition is a good place to see computer vision work end to end, and this repo walks through it with one CNN. A trained model already sorts hand-drawn doodles into 20 everyday classes from Google's Quick, Draw! dataset, and the training scripts are included. Two runnable apps make the results tangible: one follows a blue, red, or green pen in front of your webcam while you draw in the air, the other lets you draw on a canvas. Changing the category list is just editing a constant in the config and downloading the matching npz files, so you can retrain on your own set of sketches without touching the rest of the pipeline. It feels more like a compact tutorial than a production service, which makes it a practical study project for anyone getting into sketch or gesture classification.

Articles for this project

No articles for this project yet.

To suggest a topic or contribute an article, contact us.

Related projects in this category