Awesome
Deep Image Blending
This is a Pytorch implementation of our paper "Deep Image Blending".
Deep Image Blending <br /> Lingzhi Zhang, Tarmily Wen, Jianbo Shi <br /> GRASP Laboratory, University of Pennsylvania
In Winter Conference on Applications of Computer Vision (WACV), 2020
Introduction
We propose a Poisson blending loss that achieves the same purpose of Poisson Image Editing. We jointly optimize the proposed Poisson blending loss with style and content loss computed from a deep network, and reconstruct the blending region by iteratively updating the pixels using the L-BFGS solver. In the blending image, we not only smooth out gradient domain of the blending boundary but also add consistent texture into the blending region.
<img src='demo_imgs/first_demo.png' align="middle" width=540>Usage
This project uses poetry to manage dependencies; start by install poetry and then dependencies
pip install poetry
poetry install
Once this is done you can run the example. Please check the arguments in the code for you application.
poetry run python run.py
# check arguments
poetry run python run.py --help
Ablation Study
<img src='demo_imgs/ablation_study.png' align="middle" width=720>Example results for paintings
<img src='demo_imgs/painting_comparison.png' align="middle" width=720>Example results for real-world images
<img src='demo_imgs/real_comparison.png' align="middle" width=720>Citation
If you use this code for your research, please cite our paper:
@inproceedings{zhang2020deep,
title={Deep Image Blending},
author={Zhang, Lingzhi and Wen, Tarmily and Shi, Jianbo},
booktitle={The IEEE Winter Conference on Applications of Computer Vision},
pages={231--240},
year={2020}
}