templeblock / mask-guided-portrait-editing-pytorch

Unofficial Mask Guided Portrait Editing w/ cGAN implementation in pytorch

Mask-Guided-Portrait-Editing-pytorch

Unofficial Mask Guided Portrait Editing w/ cGAN implementation in pytorch.

Original Repo is here!

WIP / This implementation includes my version of tunings/tweaks :)

Introduction

Mask-Guided Portrait Editing is a novel technology based on mask-guided conditional GANs, which can synthesize diverse, high-quality and controllable facial images from given masks. With the changeable input facial mask and source image, this method allows users to do high-level portrait editing.

Architecture

Local Components Parsing Network

I use a custom network for parsing face & hair components, which is different w/ the paper. Left components (eyes, mouth, nose) are parsed w/ fixed coordination.

Performance about my face & hair parsing network, The network achieves ... on Figaro1K test dataset.

More Details are here.

Local Embedding Sub-Network

Mask-Guided Generative Sub-Network

Background Fusing Sub-Network

Usage

Result

Citation

@inproceedings{gu2019mask,
  title={Mask-Guided Portrait Editing With Conditional GANs},
  author={Gu, Shuyang and Bao, Jianmin and Yang, Hao and Chen, Dong and Wen, Fang and Yuan, Lu},
  booktitle={Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition},
  pages={3436--3445},
  year={2019}
}

Author

HyeongChan Kim