Sketch colorization can be applied in many scenarios. Our project will mainly focus on the colorization of black and white comics and manga. This will make the content more immersive for the readers. It can also help artists get a feel for the colored version of their sketches.
This is an implementation of Image-to-Image Translation with Conditional Adversarial Networks. paper by Phillip Isola, Jun-Yan Zhu, Tinghui Zhou, Alexei A. Efros
You need to install
tensorflow
matplotlib
We have used Anime Sketch Colorization Pair form Kaggle. It has a total of 17.8K images, with 14.2K train and 3.5K test split. Each image is divided into two parts; the black and white sketch and the colored version of that sketch