Fast A3RL: Aesthetics-Aware Adversarial Reinforcement Learning for Image Cropping

doi:10.1109/TIP.2019.2914360

	Fast A3RL: Aesthetics-Aware Adversarial Reinforcement Learning for Image Cropping
	Li, Debang1,2 ; Wu, Huikai1,2 ; Zhang, Junge1,2 ; Huang, Kaiqi1,2,3
刊名	IEEE TRANSACTIONS ON IMAGE PROCESSING
	2019-10-01
卷号	28 期号:10 页码:5105-5120
关键词	Reinforcement learning adversarial learning image cropping
ISSN号	1057-7149
DOI	10.1109/TIP.2019.2914360
通讯作者	Huang, Kaiqi(kqhuang@nlpr.ia.ac.cn)
英文摘要	Image cropping aims at improving the quality of images by removing unwanted outer areas, which is widely used in the photography and printing industry. Most of the previous cropping methods that do not need bounding box supervision rely on the sliding window mechanism. The sliding window method results in fixed aspect ratios and limits the shape of the cropping region. Moreover, the sliding window method usually produces lots of candidates on the input image, which is very time-consuming. Motivated by these challenges, we formulate image cropping as a sequential decision-making process and propose a reinforcement learning-based framework to address this problem, namely, Fast Aesthetics-Aware Adversarial Reinforcement Learning (Fast A3RL). Particularly, the proposed method develops an aesthetics-aware reward function that is dedicated for image cropping. Similar to human's decision-making process, we use a comprehensive state representation, including both the current observation and the historical experience. We train the agent using the actor-critic architecture in an end-to-end manner. The adversarial learning process is also applied during the training stage. The proposed method is evaluated on several popular cropping datasets, in which the images are unseen during training. The experiment results show that our method achieves the state-of-the-art performance with much fewer candidate windows and much less time compared with related methods.
资助项目	National Key Research and Development Program of China[2016YFB1001004] ; National Key Research and Development Program of China[2016YFB1001005] ; National Natural Science Foundation of China[61876181] ; National Natural Science Foundation of China[61673375] ; National Natural Science Foundation of China[61721004] ; Projects of Chinese Academy of Sciences[QYZDB-SSW-JSC006]
WOS研究方向	Computer Science ; Engineering
语种	英语
出版者	IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
WOS记录号	WOS:000482599100009
资助机构	National Key Research and Development Program of China ; National Natural Science Foundation of China ; Projects of Chinese Academy of Sciences
内容类型	期刊论文
源URL	[http://ir.ia.ac.cn/handle/173211/27309]
专题	中国科学院自动化研究所
通讯作者	Huang, Kaiqi
作者单位	1.Chinese Acad Sci, Inst Automat, Beijing 100190, Peoples R China 2.Univ Chinese Acad Sci, Sch Artificial Intelligence, Beijing 100049, Peoples R China 3.CAS Ctr Excellence Brain Sci & Intelligence Techn, Beijing 100190, Peoples R China
推荐引用方式 GB/T 7714	Li, Debang,Wu, Huikai,Zhang, Junge,et al. Fast A3RL: Aesthetics-Aware Adversarial Reinforcement Learning for Image Cropping[J]. IEEE TRANSACTIONS ON IMAGE PROCESSING,2019,28(10):5105-5120.
APA	Li, Debang,Wu, Huikai,Zhang, Junge,&Huang, Kaiqi.(2019).Fast A3RL: Aesthetics-Aware Adversarial Reinforcement Learning for Image Cropping.IEEE TRANSACTIONS ON IMAGE PROCESSING,28(10),5105-5120.
MLA	Li, Debang,et al."Fast A3RL: Aesthetics-Aware Adversarial Reinforcement Learning for Image Cropping".IEEE TRANSACTIONS ON IMAGE PROCESSING 28.10(2019):5105-5120.