Paper: https://arxiv.org/abs/2005.07702
This project takes on the problem of transferring the style of cartoon images to real-life photographic images by implementing previous work done by CartoonGAN. We trained a Generative Adversial Network(GAN) on over 60 000 images from works by Hayao Miyazaki at Studio Ghibli.
-
Install Anaconda from https://www.anaconda.com/
-
Install pytorch at: https://pytorch.org/get-started/locally/
-
Install dependencies:
python -m pip install tqdm pillow numpy matplotlib opencv-python
-
For predicting videos you will also need ffmpeg
All training code can be found in experiment.ipynb
Predict by running predict.py
.
Example:
python predict.py -i C:/folder/input_image.png -o ./output_folder/output_image.png
Predictions can be made on images, videos or a folder of images/videos.
Image # | Original | CartoonGAN | GANILLA | Our implementation |
---|---|---|---|---|
1 | ||||
2 | ||||
3 | ||||
4 | ||||
5 | ||||
6 | ||||
7 | ||||
8 | ||||
9 | ||||
10 | ||||
11 | ||||
12 | ||||
13 | ||||
14 | ||||
15 | ||||
16 | ||||
17 | ||||
18 | ||||
19 | ||||
20 |