We describe an algorithm for reconstructing the 3D shape of the scene and the relative pose of a number of cameras from a collection of images under the assumption that the scene ...
We present a system that constructs "implicit shape models" for classes of rigid 3D objects and utilizes these models to estimating the pose of class instances in single...
Many museum and library archives are digitizing their large collections of handwritten historical manuscripts to enable public access to them. These collections are only available...
Abstract. We propose a model of the shape, motion and appearance of a sequence of images that captures occlusions, scene deformations, arbitrary viewpoint variations and changes in...
Jeremy D. Jackson, Anthony J. Yezzi, Stefano Soatt...
The goal of image categorization is to classify a collection of unlabeled images into a set of predefined classes to support semantic-level image retrieval. The distance measures ...