Editable Generative Adversarial Networks: Generating and Editing Faces Simultaneously

Kyungjune Baek, Duhyeon Bang, Hyunjung Shim

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

We propose a novel framework for simultaneously generating and manipulating the face images with desired attributes. While the state-of-the-art attribute editing techniques have achieved the impressive performance for creating realistic attribute effects, they only address the image editing problem, using the input image as the condition of model. Recently, several studies attempt to tackle both novel face generation and attribute editing problem using a single model. However, their image quality is still unsatisfactory. Our goal is to develop a single unified model that can simultaneously create and edit high quality face images with desired attributes. A key idea of our work is that we decompose the image into the latent and attribute vector in low dimensional representation, and then utilize the GANs framework for mapping the low dimensional representation to the image. In this way, we can address both the generation and editing problem by training the proposed GANs, namely Editable GAN. For qualitative and quantitative evaluations, the proposed GANs outperform recent algorithms addressing the same problem. Also, we show that our model can achieve the competitive performance with the state-of-the-art attribute editing technique in terms of attribute editing quality.

Original languageEnglish
Title of host publicationComputer Vision – ACCV 2018 - 14th Asian Conference on Computer Vision, Revised Selected Papers
EditorsHongdong Li, C.V. Jawahar, Greg Mori, Konrad Schindler
PublisherSpringer Verlag
Pages39-55
Number of pages17
ISBN (Print)9783030208868
DOIs
Publication statusPublished - 2019 Jan 1
Event14th Asian Conference on Computer Vision, ACCV 2018 - Perth, Australia
Duration: 2018 Dec 22018 Dec 6

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume11361 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference14th Asian Conference on Computer Vision, ACCV 2018
CountryAustralia
CityPerth
Period18/12/218/12/6

Fingerprint

Attribute
Face
Image quality
Quantitative Evaluation
Model
Image Quality
Decompose

All Science Journal Classification (ASJC) codes

  • Theoretical Computer Science
  • Computer Science(all)

Cite this

Baek, K., Bang, D., & Shim, H. (2019). Editable Generative Adversarial Networks: Generating and Editing Faces Simultaneously. In H. Li, C. V. Jawahar, G. Mori, & K. Schindler (Eds.), Computer Vision – ACCV 2018 - 14th Asian Conference on Computer Vision, Revised Selected Papers (pp. 39-55). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11361 LNCS). Springer Verlag. https://doi.org/10.1007/978-3-030-20887-5_3
Baek, Kyungjune ; Bang, Duhyeon ; Shim, Hyunjung. / Editable Generative Adversarial Networks : Generating and Editing Faces Simultaneously. Computer Vision – ACCV 2018 - 14th Asian Conference on Computer Vision, Revised Selected Papers. editor / Hongdong Li ; C.V. Jawahar ; Greg Mori ; Konrad Schindler. Springer Verlag, 2019. pp. 39-55 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{c3d40cbab43448d4b6332bbafa7c349f,
title = "Editable Generative Adversarial Networks: Generating and Editing Faces Simultaneously",
abstract = "We propose a novel framework for simultaneously generating and manipulating the face images with desired attributes. While the state-of-the-art attribute editing techniques have achieved the impressive performance for creating realistic attribute effects, they only address the image editing problem, using the input image as the condition of model. Recently, several studies attempt to tackle both novel face generation and attribute editing problem using a single model. However, their image quality is still unsatisfactory. Our goal is to develop a single unified model that can simultaneously create and edit high quality face images with desired attributes. A key idea of our work is that we decompose the image into the latent and attribute vector in low dimensional representation, and then utilize the GANs framework for mapping the low dimensional representation to the image. In this way, we can address both the generation and editing problem by training the proposed GANs, namely Editable GAN. For qualitative and quantitative evaluations, the proposed GANs outperform recent algorithms addressing the same problem. Also, we show that our model can achieve the competitive performance with the state-of-the-art attribute editing technique in terms of attribute editing quality.",
author = "Kyungjune Baek and Duhyeon Bang and Hyunjung Shim",
year = "2019",
month = "1",
day = "1",
doi = "10.1007/978-3-030-20887-5_3",
language = "English",
isbn = "9783030208868",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer Verlag",
pages = "39--55",
editor = "Hongdong Li and C.V. Jawahar and Greg Mori and Konrad Schindler",
booktitle = "Computer Vision – ACCV 2018 - 14th Asian Conference on Computer Vision, Revised Selected Papers",
address = "Germany",

}

Baek, K, Bang, D & Shim, H 2019, Editable Generative Adversarial Networks: Generating and Editing Faces Simultaneously. in H Li, CV Jawahar, G Mori & K Schindler (eds), Computer Vision – ACCV 2018 - 14th Asian Conference on Computer Vision, Revised Selected Papers. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 11361 LNCS, Springer Verlag, pp. 39-55, 14th Asian Conference on Computer Vision, ACCV 2018, Perth, Australia, 18/12/2. https://doi.org/10.1007/978-3-030-20887-5_3

Editable Generative Adversarial Networks : Generating and Editing Faces Simultaneously. / Baek, Kyungjune; Bang, Duhyeon; Shim, Hyunjung.

Computer Vision – ACCV 2018 - 14th Asian Conference on Computer Vision, Revised Selected Papers. ed. / Hongdong Li; C.V. Jawahar; Greg Mori; Konrad Schindler. Springer Verlag, 2019. p. 39-55 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11361 LNCS).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - Editable Generative Adversarial Networks

T2 - Generating and Editing Faces Simultaneously

AU - Baek, Kyungjune

AU - Bang, Duhyeon

AU - Shim, Hyunjung

PY - 2019/1/1

Y1 - 2019/1/1

N2 - We propose a novel framework for simultaneously generating and manipulating the face images with desired attributes. While the state-of-the-art attribute editing techniques have achieved the impressive performance for creating realistic attribute effects, they only address the image editing problem, using the input image as the condition of model. Recently, several studies attempt to tackle both novel face generation and attribute editing problem using a single model. However, their image quality is still unsatisfactory. Our goal is to develop a single unified model that can simultaneously create and edit high quality face images with desired attributes. A key idea of our work is that we decompose the image into the latent and attribute vector in low dimensional representation, and then utilize the GANs framework for mapping the low dimensional representation to the image. In this way, we can address both the generation and editing problem by training the proposed GANs, namely Editable GAN. For qualitative and quantitative evaluations, the proposed GANs outperform recent algorithms addressing the same problem. Also, we show that our model can achieve the competitive performance with the state-of-the-art attribute editing technique in terms of attribute editing quality.

AB - We propose a novel framework for simultaneously generating and manipulating the face images with desired attributes. While the state-of-the-art attribute editing techniques have achieved the impressive performance for creating realistic attribute effects, they only address the image editing problem, using the input image as the condition of model. Recently, several studies attempt to tackle both novel face generation and attribute editing problem using a single model. However, their image quality is still unsatisfactory. Our goal is to develop a single unified model that can simultaneously create and edit high quality face images with desired attributes. A key idea of our work is that we decompose the image into the latent and attribute vector in low dimensional representation, and then utilize the GANs framework for mapping the low dimensional representation to the image. In this way, we can address both the generation and editing problem by training the proposed GANs, namely Editable GAN. For qualitative and quantitative evaluations, the proposed GANs outperform recent algorithms addressing the same problem. Also, we show that our model can achieve the competitive performance with the state-of-the-art attribute editing technique in terms of attribute editing quality.

UR - http://www.scopus.com/inward/record.url?scp=85066793990&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85066793990&partnerID=8YFLogxK

U2 - 10.1007/978-3-030-20887-5_3

DO - 10.1007/978-3-030-20887-5_3

M3 - Conference contribution

AN - SCOPUS:85066793990

SN - 9783030208868

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 39

EP - 55

BT - Computer Vision – ACCV 2018 - 14th Asian Conference on Computer Vision, Revised Selected Papers

A2 - Li, Hongdong

A2 - Jawahar, C.V.

A2 - Mori, Greg

A2 - Schindler, Konrad

PB - Springer Verlag

ER -

Baek K, Bang D, Shim H. Editable Generative Adversarial Networks: Generating and Editing Faces Simultaneously. In Li H, Jawahar CV, Mori G, Schindler K, editors, Computer Vision – ACCV 2018 - 14th Asian Conference on Computer Vision, Revised Selected Papers. Springer Verlag. 2019. p. 39-55. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-030-20887-5_3