Instance-level salient object segmentation

Image saliency detection has recently achieved great success due to the development of deep convolutional neural networks . However, most existing salient object methods cannot identify individual instances in detected region In this paper, we present a instance segmentation method that produces map with distinct labels for an input image. Our consists three primary steps, i.e. , inference, contours detection, and identification. For first two propose multiscale refinement network, which generates high-quality masks contours. last step, morphology algorithm incorporates regions generate promising results. To promote further research evaluation segmentation, also construct new database (ILSO-2K) 2,000 images pixel-wise annotations. Experimental results demonstrate our proposed is capable achieving satisfactory performance over six public benchmarks as well on dataset segmentation. The source code will be available at https://github.com/Kinpzz/MSRNet-CVIU • A network (MSRNet) detection. MSRNet applicable instance-level We create challenging