CycleSegNet: Object Co-Segmentation With Cycle Refinement and Region Correspondence

Image co-segmentation is an active computer vision task that aims to segment the common objects from a set of images. Recently, researchers design various learning-based algorithms undertake task. The main difficulty in this how effectively transfer information between images make conditional predictions. In paper, we present CycleSegNet, novel framework for Our network has two key components: region correspondence module which basic operation exchanging local image regions, and cycle refinement module, utilizes ConvLSTMs progressively update representations exchange iterative manner. Extensive experiments demonstrate our proposed method significantly outperforms state-of-the-art methods on four popular benchmark datasets - PASCAL VOC dataset, MSRC Internet iCoseg by 2.6%, 7.7%, 2.2%, 2.9%, respectively.