This product is an open source software. If you would like to access the source code, please visit the below link. (You do not need to click "ADD TO CART" for this item)

<<To ACCESS Source Code: CLICK HERE>>

Authors: Yingqing He, Yazhou Xing, Tianjia Zhang, Qifeng Chen

Portrait images often suffer from undesirable shadows cast by casual objects or even the face itself. While existing methods for portrait shadow removal require training on a large-scale synthetic dataset, we propose the first unsupervised method for portrait shadow removal without any training data. Our key idea is to leverage the generative facial priors embedded in the off-the-shelf pretrained StyleGAN2. To achieve this, we formulate the shadow removal task as a layer decomposition problem: a shadowed portrait image is constructed by the blending of a shadow image and a shadow-free image. We propose an effective progressive optimization algorithm to learn the decomposition process. Our approach can also be extended to portrait tattoo removal and watermark removal. Qualitative and quantitative experiments on a real-world portrait shadow dataset demonstrate that our approach achieves comparable performance with supervised shadow removal methods.

Authors:

Department of Computer Science and Engineering Assistance Professor Prof. Qifeng Chen

Key Features:

We proposed the first unsupervised method for portrait shadow removal which needs only one input shadow portrait image. We have shown that the generative priors can be used in this unsupervised layer decomposition setting to handle unknown degradation processes which cannot be accomplished by existing GAN-inversion methods. Meanwhile, we designed progressive optimization techniques to guide the image decomposition and reconstruction process. Then, we achieved comparable performance with existing state-of-the-art supervised-based shadow removal methods, demonstrating the effectiveness of our unsupervised method. Finally, we have shown two extension applications (e.g., portrait tattoo removal and watermark removal) of our method to demonstrate that our method can serve as a unified framework for portrait image decomposition tasks.

Specification:

Please refer to the published paper for specification

Reference:

https://cqf.io/papers/Portrait_Shadow_Removal_ACMMM2021.pdf

Contact Us:

ttsamuel@ust.hk

Download (1)

Unsupervised Portrait Shadow Removal via Generative Priors
Portrait_Shadow_Removal_ACMMM2021.pdf*
size: 5646013 KB, type: application/pdf
 

You would be able to download the files marked with an asterisk '*' only after logging in to your account and completing the purchase of the product license.