Image cropping aims at improving the aesthetic quality of images by adjustingtheir composition. Most previous methods rely on the sliding window mechanism.The sliding window mechanism requires fixed aspect ratios and limits thecropping region with arbitrary size. Moreover, the sliding window methodusually produces tens of thousands of windows which is very time-consuming.Motivated by these challenges, we firstly formulate the aesthetic imagecropping as a sequential decision-making process and propose an AestheticsAware Reinforcement Learning (A2-RL) framework to address this problem.Particularly, the proposed method develops an aesthetics aware reward functionwhich especially benefits image cropping. Similar to human's decision making,we use a comprehensive state representation including both the currentobservation and the historical experience. We train the agent using theactor-critic architecture in an end-to-end manner. The agent is evaluated onseveral popular unseen cropping datasets. Experiment results show that ourmethod achieves the state-of-the-art performance with much fewer candidatewindows and much less time compared with previous methods.
展开▼
