FocusGAN: Preserving Background in Text-Guided Image Editing
-
Published:2021-12-20
Issue:
Volume:
Page:
-
ISSN:0218-0014
-
Container-title:International Journal of Pattern Recognition and Artificial Intelligence
-
language:en
-
Short-container-title:Int. J. Patt. Recogn. Artif. Intell.
Author:
Zhao Liuqing1,
Li Linyan2,
Hu Fuyuan1,
Xia Zhenping1,
Yao Rui3
Affiliation:
1. School of Electronic & Information Engineering, Suzhou University of Science and Technology, Suzhou 215000, P. R. China
2. School of Information Technology, School of Suzhou Institute of Trade & Commerce, Suzhou 215000, P. R.China
3. School of Computer Science and Technology China University of Mining and Technology Xuzhou 221000, P. R. China
Abstract
Text-guided image editing (TIE) seeks to manipulate images by the guidance of language. However, the existing TIE methods always overlook the target-irrelevant pixels and the editing may make the background discolored, distorted, or partially disappear. To overcome this problem, we propose a novel TIE method named FocusGAN, which edits the text-relevant pixels precisely as well as keeps the background invariant. Specifically, we build a network of two stages. In each stage, we first construct a channel-wise subject focusing attention to make the generator focus on the sub-region that best matches each word. Then, the word-level discriminator provides fine-grained feedback by correlating words with image areas, so that the generator can manipulate specific visual attributes without affecting the background. Last, we propose a background-keeping cyclic loss to further improve the invariance of the background and to encourage the edit of the subject that matches the given text. Experiments on CUB and Oxford datasets demonstrate that our approach can effectively keep the background invariant in manipulating images using natural language descriptions.
Funder
Natural Science Foundation of China
Jiangsu Provincial Key Research and Development Program
Scientific Research Project of School of Suzhou Institute of Trade & Commerce
Natural Science Foundation of Jiangsu Province
Publisher
World Scientific Pub Co Pte Ltd
Subject
Artificial Intelligence,Computer Vision and Pattern Recognition,Software
Cited by
1 articles.
订阅此论文施引文献
订阅此论文施引文献,注册后可以免费订阅5篇论文的施引文献,订阅后可以查看论文全部施引文献