For AIM2020 ECCV Extreme Image Inpainting Challenge (both Track 1 Classic and Track 2 Semantic Guidance)
This is the Pytorch implementation of DeepGIN for Extreme Image Inpainting. We have participated in AIM 2020 ECCV Extreme Image Inpainting Challenge. DeepGIN is used for reconstructing a completed image from a randomly masked image with both satisfactory visual quality and pixel-wise reconstruction accuracy.
For more information about the challenge, please visit the github project page provided by the organizers here. Thank you very much!!
- For Track 1 Classic Inpainting, please click here (our project link to track 1)
- For Track 2 Inpainting with Semantic Guidance, please click here (our project link to track 2)
Thanks for visiting our project page, if it is useful, please cite our paper,
@misc{li2020deepgin,
title={DeepGIN: Deep Generative Inpainting Network for Extreme Image Inpainting},
author={Chu-Tak Li and Wan-Chi Siu and Zhi-Song Liu and Li-Wen Wang and Daniel Pak-Kong Lun},
year={2020},
eprint={2008.07173},
archivePrefix={arXiv},
primaryClass={cs.CV}
}
- Thank you for visiting this project page
- Our code is developed based on the skeleton of the Pytorch implementation of pix2pixHD