This repo is the final project of lecture 3dcv.
Automatic image colorization has made significant advancements with the advent of deep learning techniques. Against this background, this project provides a comprehensive comparison of three cutting-edge image colorization methods: Convolutional Neural Networks (CNNs), Generative Adversarial Networks (GANs), and Diffusion Models. We implemented the three mentioned models and delved into the foundational principles of each method, outlining our implementational decisions.
The environment needed can be built via:
conda env create -f environment.yml
The train.py contains detailed help list which enables you set various training parameters.
python train.py --help
The demo of our project is available in demo.ipynb.
We use AFHQ cat dataset and use dataset_tool.py to compress the images to 256*256.
The dataset used in our experiments can be downloaded via:
https://drive.google.com/file/d/1qd2fSfJp2-o06S-kJgildS1PBrj353AY/view?usp=sharing