We introduce an approach to accurately detect and segment partially occluded objects in various viewpoints and scales. Our main contribution is a novel framework for combining obj...
A new approach for separating mathematics from usual text is presented. Contrary to the existing methods, it is more oriented toward the segmentation than the recognition, isolati...
Active musicians generally rely on extensive personal paper-based music information retrieval systems containing scores, parts, compositions, and arrangements of published and han...
Efficient detection of objects in images is complicated by variations of object appearance due to intra-class object differences, articulation, lighting, occlusions, and aspect va...
Histograms of local appearance descriptors are a popular representation for visual recognition. They are highly discriminant with good resistance to local occlusions and to geomet...