Automatically colorize images using Machine Learning.

• Project page
• ECCV 2016 paper (arXiv)

Make sure that you have Caffe (with Python bindings). Then run:

pip install autocolorize

Run:

autocolorize grayscale.png -o colorized.png

You can also colorize from Python (assuming grayscale is the image that you want to colorize):

import autocolorize
classifier = autocolorize.load_default_classifier()
rgb = autocolorize.colorize(grayscale, classifier=classifier)

We provide custom layers for doing sparse hypercolumn training in both Caffe (see caffe/) and Tensorflow (see tensorflow/). This can be used for other image-to-image tasks, such as semantic segmentation or edge prediction.

Look inside the train folder if you want to train from scratch.