Does Image Segmentation Improve Object Categorization?

Image segmentation and object recognition are among the most fundamental problems in computer vision, and the potential interaction between these tasks has been discussed for many years. The usefulness of recognition for segmentation has been demonstrated with various top-down segmentation algorithms, however, the impact of bottom-up image segmentation as pre-processing for object recognition is not well understood. One factor impeding the utility of segmentation for recognition is the unsatisfactory quality of image segmentation algorithms. In this work we take advantage of a recently proposed method for computing multiple stable segmentations and illustrate the application of bottom-up image segmentation as a preprocessing step for object recognition and categorization. We extend a popular bag-of-features recognition model to provide multiple class categorization and localization of objects in images. We compare our categorization results to that of a conventional bag-of-features recognition model on the Caltech and PASCAL image databases.