Dual-Branch Network of Information Mutual Optimization for Salient Object Detection

Salient object detection (SOD) is to segment significant regions of images. Noticing that the saliency maps in existing SOD methods suffer from blurring boundaries owing insufficient extraction boundary features and inadequate fusion between salient region features, a dual-branch network information mutual optimization (DIMONet) proposed. The DIMONet has branch extract corresponding simultaneously mainly composed two components. One module (MOM) refines based on their internal relationship. other multi-receptive fields (FMMF) integrates multi-layer with refined distinguish objects better sharpen boundaries. With help MOMs FMMFs, noises background are gradually reduced hence get sharpened. Experiments five benchmark datasets show our method superior 18 state-of-the-art methods.