Abstract

This paper presents a nonparametric scene parsing approach that improves the overall accuracy, as well as the coverage of foreground classes in scene images. We fifirst improve the label likelihood estimates at superpixels by merging likelihood scores from different probabilistic classififiers. This boosts the classifification performance and enriches the representation of less-represented classes. Our second contribution consists of incorporating semantic context in the parsing process through global label costs. Our method does not rely on image retrieval sets but rather assigns a global likelihood estimate to each label, which is plugged into the overall energy function. We evaluate our system on two large-scale datasets, SIFTflflow and LMSun. We achieve state-of-the-art performance on the SIFTflflow dataset and near-record results on LMSun