Active contours or snakes have been extensively utilized in handling image segmentation and classification problems. In traditional active contour models, snake initialization is ...
We present a random field based model for stereo vision with explicit occlusion labeling in a probabilistic framework. The model employs non-parametric cost functions that can be ...
We apply Stochastic Meta-Descent (SMD), a stochastic gradient optimization method with gain vector adaptation, to the training of Conditional Random Fields (CRFs). On several larg...
S. V. N. Vishwanathan, Nicol N. Schraudolph, Mark ...
This paper describes a new algorithm for computing linear generators (vector generating polynomials) for matrix sequences, running in subquadratic time. This algorithm applies in ...
—We introduce a robust image segmentation method based on a variational formulation using edge flow vectors. We demonstrate the nonconservative nature of this flow field, a fe...
Pratim Ghosh, Luca Bertelli, Baris Sumengen, B. S....