CO-SALIENCY OBJECT DETECTION BASED ON INTRA-GROUP CONSISTENCY AND COMPARATIVE LEARNING

Aimed at the problem of false detection of co-salient object and imprecise object boundaries in co-saliency object detection, a co-saliency object detection algorithm based on intra-group consistency and comparative learning is proposed. The intra-group consistency module was designed to explore the common attributes of the image group, and the contrast learning was introduced to improve the differentiation of the common attributes to different salient objects, which improved the detection accuracy. The feature fusion module was designed to optimize the boundary of salient objects in the process of multi-scale feature fusion, which improved the segmentation effect. Experiments on three benchmark datasets show that the performance of this algorithm is better than the current mainstream algorithms.