Deep reinforcement learning in computer vision: a comprehensive survey

Deep reinforcement learning augments the framework and utilizes powerful representation of deep neural networks. Recent works have demonstrated remarkable successes in various domains including finance, medicine, healthcare, video games, robotics, computer vision. In this work, we provide a detailed review recent state-of-the-art research advances We start with comprehending theories learning, learning. then propose categorization methodologies discuss their advantages limitations. particular, divide into seven main categories according to applications vision, i.e. (i) landmark localization (ii) object detection; (iii) tracking; (iv) registration on both 2D image 3D volumetric data (v) segmentation; (vi) videos analysis; (vii) other applications. Each these is further analyzed techniques, network design, performance. Moreover, comprehensive analysis existing publicly available datasets examine source code availability. Finally, present some open issues future directions