Given an input image, our method CBS can stylize a certain object class with a specifi style. Run the following command to achieve a result similar to the one below.
python main.py -i images/example.png -o images/stylized.png -s 0 -c 13
In this work we combined these two pretrained models to segment different objects in a scene. But if you would like to train for segmenting new objects or new styles, you can use the training scripts in [1,2].
[1] https://github.com/Reagan1311/DABNet/
[2] https://github.com/pytorch/examples/tree/master/fast_neural_style
If you find the code useful for your research, please cite:
@inproceedings{kurzman2019class,
title={Class-Based Styling: Real-time Localized Style Transfer with Semantic Segmentation},
author={Kurzman, Lironne and Vazquez, David and Laradji, Issam},
booktitle={Proceedings of the IEEE International Conference on Computer Vision Workshops},
pages={0--0},
year={2019}
}