Awesome
SketchEdit: Mask-Free Local Image Manipulation with Partial Sketches
[Paper] [Project Page] [Interactive Demo] [Supplementary Material]
Update
Star this project to get notified.
<p align="center"> <a href="#"><img src="https://user-images.githubusercontent.com/11280511/150838003-01136487-6671-4cc4-ae4d-bef8e60c605e.png"></a> </p>-
:new: Uploading custom photos are now enabled in the demo
-
:new: Edit face in a fullbody/upper body photo
<img src="https://github.com/zengxianyu/sketchedit/blob/main/example_body.jpg" width=640> -
:new: Enable general scenes editing in the demo
-
Exposing predicted mask & mask revising?
Usage
environment
The code only has been tested on Ubuntu 20.04.
GPU is required.
- Clone the repo:
git clone --single-branch https://github.com/zengxianyu/sketchedit
git submodule init
git submodule update
- Install using the provided docker file
Dockerfile
or
conda env create -f environment.yml
- Switch to the installed environment
conda activate editline
Testing
- Download pretrained model
chmod +x download/*
./download/download_model.sh
- Use the model trained on CelebAHQ
./test_celeb.sh
- Use the model trained on Places
./test_places.sh
Interactive demo
Training code and data
Training code and data is coming soon. Star this project to get notified.
<p align="center"> <a href="#"><img src="https://user-images.githubusercontent.com/11280511/150838003-01136487-6671-4cc4-ae4d-bef8e60c605e.png"></a> </p>Citation
@inproceedings{zeng2022sketchedit,
title={SketchEdit: Mask-Free Local Image Manipulation with Partial Sketches},
author={Zeng, Yu and Lin, Zhe and Patel, Vishal M.},
booktitle={Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition},
year={2022}
}