A Mutual Learning Method for Salient Object Detection with intertwined Multi-Supervision

Though deep learning techniques have made great progress in salient object detection recently, the predicted saliency maps still suffer from incomplete predictions due to the internalcomplexity of objects and inaccurateboundaries caused by strides in convolution and pooling operations. To alleviate these issues, we propose to train saliency detection networks by exploiting the supervisionfrom not only salient object detection, but alsoforeground contour detection and edge detection. First,we leverage salient object detection andforeground contour detection tasks in an intertwinedmanner to generate saliency maps with uniform highlight. Second, the foregroundcontour and edge detection tasks guide each othersimultaneously, thereby leading to precise foreground contourprediction and reducing the local noises for edge prediction. In addition, we develop a novel mutual learning module (MLM) which serves as the building block of our method. Each MLM consists of multiple network branches trained in a mutual learning manner which improves the performance by a large margin. Extensive experiments on seven challengingdatasetsdemonstrate that the proposed method has delivered state-of-the-art results in both salient object detection and edge detection.