Abstract

In this paper, we propose a novel method to perform weakly-supervised image parsing based on the dictionary learning framework. To deal with the challenges caused by the label ambiguities, we design a saliency guided weight assignment scheme to boost the discriminative dictionary learning. More specififically, with a collection of tagged images, the proposed method fifirst conducts saliency detection and automatically infers the confifidence for each semantic class to be foreground or background. These clues are then incorporated to learn the dictionaries, the weights, as well as the sparse representation coeffificients in the meanwhile. Once obtained the coeffificients of a superpixel, we use a sparse representation classififier to determine its semantic label. The approach is validated on the MSRC21, PASCAL VOC07, and VOC12 datasets. Experimental results demonstrate the encouraging performance of our approach in comparison with some state-of-the-arts.