Guided Co-Segmentation Network for Fast Video Object Segmentation

Semi-supervised video object segmentation is a task of propagating instance masks given in the first frame to entire video. It challenging since it usually suffers from heavy occlusions, large deformation, and variations objects. To alleviate these problems, many existing works apply time-consuming techniques such as fine-tuning, post-processing, or extracting optical flow, which makes them intractable for online segmentation. In our work, we focus on semi-supervised We propose GCSeg (Guided Co-Segmentation) Network mainly composed Reference Module Co-segmentation Module, simultaneously incorporate short-term, middle-term, long-term temporal inter-frame relationships. Moreover, an Adaptive Search Strategy reduce risk inaccurate results subsequent frames. Our network achieves state-of-the-art performance Davis 2016 2017 datasets.