Whereas most existing action recognition methods require computationally demanding feature extraction and/or classification, this paper presents a novel real-time solution that ut...
3D object detection and importance regression/ranking are at the core for semantically interpreting 3D medical images of computer aided diagnosis (CAD). In this paper, we propose ...
Le Lu, Jinbo Bi, Matthias Wolf, Marcos Salganicoff
Human motion can be seen as a type of texture pattern. In this paper, we adopt the ideas of spatiotemporal analysis and the use of local features for motion description. movements...
—Optical flow algorithms are difficult to apply to robotic vision applications in practice because of their extremely high computational and frame rate requirements. In most case...
Epipolar geometry and relative camera pose computation are examples of tasks which can be formulated as minimal problems and solved from a minimal number of image points. Finding ...