Awesome
GP-GAN: Towards Realistic High-Resolution Image Blending (ACMMM 2019, oral)
[Project] [Paper] [Demo] [Related Work: A2RL (for Auto Image Cropping)] [Colab]
Official Chainer implementation of GP-GAN: Towards Realistic High-Resolution Image Blending
Overview
source | destination | mask | composited | blended |
---|---|---|---|---|
The author's implementation of GP-GAN, the high-resolution image blending algorithm described in:
"GP-GAN: Towards Realistic High-Resolution Image Blending"
Huikai Wu, Shuai Zheng, Junge Zhang, Kaiqi Huang
Given a mask, our algorithm can blend the source image and the destination image, generating a high-resolution and realsitic blended image. Our algorithm is based on deep generative models Wasserstein GAN.
Contact: Hui-Kai Wu (huikaiwu@icloud.com)
Citation
@article{wu2017gp,
title = {GP-GAN: Towards Realistic High-Resolution Image Blending},
author = {Wu, Huikai and Zheng, Shuai and Zhang, Junge and Huang, Kaiqi},
journal = {ACMMM},
year = {2019}
}
Getting started
-
The code is tested with
python==3.5
andchainer==6.3.0
onUbuntu 16.04 LTS
. -
Download the code from GitHub:
git clone https://github.com/wuhuikai/GP-GAN.git cd GP-GAN
-
Install the requirements:
pip install -r requirements/test/requirements.txt
-
Download the pretrained model
blending_gan.npz
orunsupervised_blending_gan.npz
from Google Drive, and then put them in the foldermodels
. -
Run the script for
blending_gan.npz
:python run_gp_gan.py --src_image images/test_images/src.jpg --dst_image images/test_images/dst.jpg --mask_image images/test_images/mask.png --blended_image images/test_images/result.png
Or run the script for
unsupervised_blending_gan.npz
:python run_gp_gan.py --src_image images/test_images/src.jpg --dst_image images/test_images/dst.jpg --mask_image images/test_images/mask.png --blended_image images/test_images/result.png --supervised False
-
Type
python run_gp_gan.py --help
for a complete list of the arguments.
Train GP-GAN step by step
Train Blending GAN
-
Download Transient Attributes Dataset here.
-
Crop the images in each subfolder:
python crop_aligned_images.py --data_root [Path for imageAlignedLD in Transient Attributes Dataset]
-
Train Blending GAN:
python train_blending_gan.py --data_root [Path for cropped aligned images of Transient Attributes Dataset]
-
Training Curve
-
Visual Result
Training Set Validation Set
Training Unsupervised Blending GAN
-
Requirements
pip install git+git://github.com/mila-udem/fuel.git@stable
-
Download the hdf5 dataset of outdoor natural images: ourdoor_64.hdf5 (1.4G), which contains 150K landscape images from MIT Places dataset.
-
Train unsupervised Blending GAN:
python train_wasserstein_gan.py --data_root [Path for outdoor_64.hdf5]
-
Training Curve
-
Samples after training
Visual results
Mask | Copy-and-Paste | Modified-Poisson | Multi-splines | Supervised GP-GAN | Unsupervised GP-GAN |
---|---|---|---|---|---|