Batch process implementation for the grayscale image colorization

Objective

The purpose of this project is to create a batch process implementation of the graysace image colorization process created by Richard Zhang, Phillip Isola and Alexei A. Efros (http://richzhang.github.io/colorization/). The Colorful Image Colorization project is using Caffe to colorize images using a trained network. The installation of Caffe and dependencies is a complex process and cannot be always done on every environment easily. This project is using docker to simplify the process and make it work on any environment (tested on Centos 7) with docker installed.

Prerequisites

For the process to work docker is needed. You can install Docker and configure it to start on boot on Centos 7 using folowing commands:

sudo yum install docker -y
sudo chkconfig docker on
sudo service docker start

To do it on other Linux flavours you have to use specific commands for the version you are running.

Installation

Clone this repository to a local folder. For example in the home folder.

cd ~
git clone -b master --single-branch https://github.com/sergekatzmann/batch_colorize.git

Optional

To prepare the docker image you can pull the image from the docker repository or build it yourself.

To build the image perform the following command:

cd ~/batch_colorize
sudo bash build.sh

To pull from docker repo use this command:

sudo docker pull sergekatzmann/batch_colorize:latest

Image preparation

Switch into the project folder and then go to images/in folder and place your grayscale images in this location. Thats it for this part.

You can perform the copy by something similar to the following lines:

Replace <image location> with the source location of your images.

cd ~/batch_colorize/images/in
cp <image location> .

Run the batch colorization

Perform the following command to colorize all the images from the images/in folder:

sudo bash batch_colorize.sh

The results

After the successful execution you will find the colorized images in the images/out folder.

Thanks

Thanks to Richard Zhang, Phillip Isola and Alexei A. Efros for the great project making alot of people happy. At least my family is very happy to see the old photos in the new colorful version.