The original paper is in English. Non-English content has been machine-translated and may contain typographical errors or mistranslations. ex. Some numerals are expressed as "XNUMX".
Copyrights notice
The original paper is in English. Non-English content has been machine-translated and may contain typographical errors or mistranslations. Copyrights notice
A síntese de esboço facial refere-se à transformação de fotos faciais em esboços. Pesquisas recentes sobre síntese de esboços faciais alcançaram grande sucesso devido ao desenvolvimento de Redes Adversariais Generativas (GAN). No entanto, estes métodos generativos tendem a negligenciar informações detalhadas e, assim, perder algumas características específicas individuais, como óculos e cocares. Neste artigo, propomos um novo método chamado Feature Learning Generative Adversarial Network (FL-GAN) para sintetizar esboços de alta qualidade com preservação de detalhes. Precisamente, o FL-GAN proposto consiste em um módulo de Feature Learning (FL) e um módulo de Adversarial Learning (AL). O módulo FL visa aprender as informações detalhadas da imagem em um espaço latente e orientar o módulo AL para sintetizar o esboço que preserva os detalhes. O Módulo AL visa aprender a estrutura e textura do esboço e melhorar a qualidade do esboço sintético por meio de estratégia de aprendizagem adversária. Comparações quantitativas e qualitativas com sete métodos de última geração, como o LLE, o MRF, o MWF, o RSLCR, o RL, o FCN e o GAN em quatro conjuntos de dados de esboço facial demonstram a superioridade deste método.
Lin CAO
Beijing Information Science and Technology University
Kaixuan LI
Beijing Information Science and Technology University
Kangning DU
Beijing Information Science and Technology University
Yanan GUO
Beijing Information Science and Technology University
Peiran SONG
Beijing Information Science and Technology University
Tao WANG
Beijing Information Science and Technology University
Chong FU
Northeastern University
The copyright of the original papers published on this site belongs to IEICE. Unauthorized use of the original or translated papers is prohibited. See IEICE Provisions on Copyright for details.
Copiar
Lin CAO, Kaixuan LI, Kangning DU, Yanan GUO, Peiran SONG, Tao WANG, Chong FU, "FL-GAN: Feature Learning Generative Adversarial Network for High-Quality Face Sketch Synthesis" in IEICE TRANSACTIONS on Fundamentals,
vol. E104-A, no. 10, pp. 1389-1402, October 2021, doi: 10.1587/transfun.2020EAP1114.
Abstract: Face sketch synthesis refers to transform facial photos into sketches. Recent research on face sketch synthesis has achieved great success due to the development of Generative Adversarial Networks (GAN). However, these generative methods prone to neglect detailed information and thus lose some individual specific features, such as glasses and headdresses. In this paper, we propose a novel method called Feature Learning Generative Adversarial Network (FL-GAN) to synthesize detail-preserving high-quality sketches. Precisely, the proposed FL-GAN consists of one Feature Learning (FL) module and one Adversarial Learning (AL) module. The FL module aims to learn the detailed information of the image in a latent space, and guide the AL module to synthesize detail-preserving sketch. The AL Module aims to learn the structure and texture of sketch and improve the quality of synthetic sketch by adversarial learning strategy. Quantitative and qualitative comparisons with seven state-of-the-art methods such as the LLE, the MRF, the MWF, the RSLCR, the RL, the FCN and the GAN on four facial sketch datasets demonstrate the superiority of this method.
URL: https://global.ieice.org/en_transactions/fundamentals/10.1587/transfun.2020EAP1114/_p
Copiar
@ARTICLE{e104-a_10_1389,
author={Lin CAO, Kaixuan LI, Kangning DU, Yanan GUO, Peiran SONG, Tao WANG, Chong FU, },
journal={IEICE TRANSACTIONS on Fundamentals},
title={FL-GAN: Feature Learning Generative Adversarial Network for High-Quality Face Sketch Synthesis},
year={2021},
volume={E104-A},
number={10},
pages={1389-1402},
abstract={Face sketch synthesis refers to transform facial photos into sketches. Recent research on face sketch synthesis has achieved great success due to the development of Generative Adversarial Networks (GAN). However, these generative methods prone to neglect detailed information and thus lose some individual specific features, such as glasses and headdresses. In this paper, we propose a novel method called Feature Learning Generative Adversarial Network (FL-GAN) to synthesize detail-preserving high-quality sketches. Precisely, the proposed FL-GAN consists of one Feature Learning (FL) module and one Adversarial Learning (AL) module. The FL module aims to learn the detailed information of the image in a latent space, and guide the AL module to synthesize detail-preserving sketch. The AL Module aims to learn the structure and texture of sketch and improve the quality of synthetic sketch by adversarial learning strategy. Quantitative and qualitative comparisons with seven state-of-the-art methods such as the LLE, the MRF, the MWF, the RSLCR, the RL, the FCN and the GAN on four facial sketch datasets demonstrate the superiority of this method.},
keywords={},
doi={10.1587/transfun.2020EAP1114},
ISSN={1745-1337},
month={October},}
Copiar
TY - JOUR
TI - FL-GAN: Feature Learning Generative Adversarial Network for High-Quality Face Sketch Synthesis
T2 - IEICE TRANSACTIONS on Fundamentals
SP - 1389
EP - 1402
AU - Lin CAO
AU - Kaixuan LI
AU - Kangning DU
AU - Yanan GUO
AU - Peiran SONG
AU - Tao WANG
AU - Chong FU
PY - 2021
DO - 10.1587/transfun.2020EAP1114
JO - IEICE TRANSACTIONS on Fundamentals
SN - 1745-1337
VL - E104-A
IS - 10
JA - IEICE TRANSACTIONS on Fundamentals
Y1 - October 2021
AB - Face sketch synthesis refers to transform facial photos into sketches. Recent research on face sketch synthesis has achieved great success due to the development of Generative Adversarial Networks (GAN). However, these generative methods prone to neglect detailed information and thus lose some individual specific features, such as glasses and headdresses. In this paper, we propose a novel method called Feature Learning Generative Adversarial Network (FL-GAN) to synthesize detail-preserving high-quality sketches. Precisely, the proposed FL-GAN consists of one Feature Learning (FL) module and one Adversarial Learning (AL) module. The FL module aims to learn the detailed information of the image in a latent space, and guide the AL module to synthesize detail-preserving sketch. The AL Module aims to learn the structure and texture of sketch and improve the quality of synthetic sketch by adversarial learning strategy. Quantitative and qualitative comparisons with seven state-of-the-art methods such as the LLE, the MRF, the MWF, the RSLCR, the RL, the FCN and the GAN on four facial sketch datasets demonstrate the superiority of this method.
ER -