Colorful

Keras implementation of Colorful Image Colorization by Richard Zhang, Phillip Isola and Alexei A. Efros

The technique is applied on the CelebA dataset with minor modifications.

Requirements

python modules

• scikit-learn==0.19.1
• scikit-image==0.13.1
• tqdm==4.17.0
• opencv_python==3.3.0.10
• numpy==1.13.3
• matplotlib==2.0.2
• Keras==2.0.8
• Theano==0.9.0 or Tensorflow==1.3.0
• h5py==2.7.0
• parmap==1.5.1
• scipy==1.0.0

System requirements

• Nvidia GPU with at least 2GB RAM
• At least 4GB RAM (when using the on_demand option for training)

The settings above should work well enough for small image size (32 x 32). Above that, better GPU and more RAM are required.

Part 1. Processing the data

Follow these instructions.

Part 2. Running the code

Follow these instructions

Part 3. Example results

For each triplet:

• First column is the original
• Second column is the B&W version
• Last column is the colorized output

Part 4. Live colorization with webcam

Follow these instructions