Fusing Global and Local Scale for Semantic Image Segmentation
Xavier Boix, Josep M. Gonfaus, Joost van de Weijer, Andrew D. Bagdanov, Joan Serrat, Jordi Gonzàlez · 2011
The Hierarchical Conditional Random Field (HCRF) model have been successfully applied to a num- ber of image labeling problems, including image segmenta- tion. However, existing HCRF models of image segmenta- tion do not allow multiple classes to be assigned to a single region, which limits their ability to incorporate contextual information across multiple scales. At higher scales in the image, this representation yields an oversimplified model since multiple classes can be reasonably expected to appear within large regions. This simplified model particularly lim- its the impact of information at higher scales. Since class- label information at these scales is usually more reliable than at lower, noisier scales, neglecting this information is undesirable. To address these issues, we propose a new con- sistency potential for image labeling problems, which we call the harmony potential. It can encode any possible com- bination of labels, penalizing only unlikely combinations of classes. We also propose an effective sampling strategy over this expanded label set that renders tractable the underlying optimization problem. Our approach obtains state-of-the- art results on two challenging, standard benchmark datasets