Image Colorization using UNet

Upload a square image. If the image is not square, it will be center-cropped to a square image before resizing to 224x224.

The input image will also be converted to the LAB color space and the L channel will be given as input to the model.

This Huggingface space is running entirely on CPU. For faster performance, consider running it locally with a GPU or use Google Colab/Kaggle notebooks.