Restricted Boltzmann Machine for saliency detection
Shijing Dong, Jinqing Qi · 2015
Saliency detection is the task of locating informative regions and objects in an image, which is a challenging task in computer vision. In this paper, we introduce an effective generative model using the Restricted Boltzmann Machine (RBM) for salient object detection. First, RBM is adopted to model the global shape of input images based on regional features. Second, an effective optimization method is used to refine the initial shape map with local relations and detailed information. Experimental results on benchmark datasets demonstrate that the proposed RBM model for saliency detection works more effectively than some existing state-of-the-art algorithms.