ARobustMultipleHypothesisApproach.ppt

ESTIMATION ALGORITHMS FOR AMBIGUOUS VISUAL MODELSReconstructing 3D Human Motion from Monocular Video Cristian Sminchisescu Goal Approach to Visual Modeling Generative Human Model Complex, kinematics, geometry, photometry Predicts images or descriptors Model-image matching cost function Associates model predictions to image features Robust, often probabilistically motivated Search/Optimization Discovers well supported configurations on resulting cost function Presentation Plan Introduction Goals, applications, general formulation Modeling (human model and image features) Parameterization, constraints and priors Error models, image descriptors State of the art and parameter estimation difficulties Methods for locating and tracking multiple minima Covariance Scaled Sampling Hyperdynamic Sampling Eigenvector Tracking Hypersurface Sweeping Kinematic Jump Sampling Human Body Model Parameter Space Priors Image Features, Integrated Robustly 2.Contours 3.Silhouettes Tracking and Optimization Formulate tracking as an optimization problem Very nonlinear kinematics, image projection, occlusion ... High dimensional search space (~35 d.o.f.) Most depth information is lost about 1/3 of d.o.f. are statically unobservable Data association is hard non-rigidity, modeling imperfections clothing, shadows, occlusion confuse the signal General unknown motions Many unexpected variables, humans switch activities, avoid obstacles Why is monocular pose hard? Tracking/Sampling Methods Local (Kalman filter, local optimization) Multiple cameras to regularize the cost surface Rehg’95, Gavrila’96 (discrete search), Kakadiaris’96, Bregler’98, Wachter’99 (1 camera), Delamarre’99, Plankers’01 Quasi-Global CONDENSATION/particle fliters (unscented, auxiliary,etc) 3D annealing, temporal models (3 cameras) (Deutscher’00), importance sampling, strong prior model (Sidenbladh’00,02) MCMC Annealing - Neal’97,’01 Tampering (Marinari’92, Neal’94) `Hybrid’ - Duane’87, Forsyth’00, Choo’0