"Real Time User-guided Colorization with Learned Deep Priors" implemented in pytorch

This is a pytorch implementation of "Real-Time User-Guided Image Colorization with Learned Deep Priors" by Zhang et.al.

Getting Started

Prerequisites

torch==0.2.0.post4, torchvision==0.1.9 The code is written with the default setting that you have gpu. Cpu mode is not recommended when using this repository.

Installing and running the tests

Make sure you have cifar10 or CelebA downloaded in ./data. You can download it through by taking a look at my "download.sh" file

./data/CelebA
./data/Cifar10
./data/pts_in_hull.npy

first clone this repository

git clone https://github.com/sjooyoo/https://github.com/sjooyoo/real-time-user-guided-colorization_pytorch.git

then run train

python deep_color.py

to sample results you first need to run deep_color.py, which will automatically save models under a models folder that will be made in your root directory. I did not include pretrained models in this repository. The --model unet100.pkl below is a sample after 100 epochs. Change the command according to your model that you want to sample.

python sampling.py --model unet100.pkl

Results

Input black and white image

Predicted colorization output

Ground truth image

Note

This is not a complete implementation. I have implemented the global hints network but have yet to incorporate it into the main network.

Further work

• global hints network

Acknowledgments

Original paper "Real-Time User-Guided Image Colorization with Learned Deep Priors"