Outline Objects using Deep Reinforcement Learning

Zhenxin Wang, Sayan Sarcar, Jingxin Liu +2

Image segmentation needs both local boundary position information and global object context information. The performance of the recent state-of-the-art method, fully convolutional…