A dynamic Bayesian network approach to figure tracking using learned dynamic models

Pavlović, Vladimir; Rehg, James M.; Cham, Tat‐Jen; Murphy, Kevin

doi:10.1109/iccv.1999.791203

Cited by 176 publications

(132 citation statements)

References 15 publications

Supporting

Mentioning

132

Contrasting

Order By: Relevance

“…Ong and Gong [112] map training data into the state space and use PCA to find a linear subspace where the training data can be compactly represented without losing too much information. Pavlović et al [115] take this idea a step farther by learning the possible or rather likely trajectories in the state space from training data; i.e., dynamic models are learned. Yet another approach is to rerepresent the state space more efficiently (without losing any information as in PCA).…”

Section: Direct Model Usementioning

confidence: 99%

“…A model of velocity and acceleration [101] or more advanced models of movements such as walking [123] may be used. An alternative approach is to learn probabilistic motion models prior to operation [115]. A commonly used method for prediction is the Kalman filter, which is also capable of estimating the uncertainties of the prediction.…”

Section: Tracking Over Timementioning

confidence: 99%

See 1 more Smart Citation

A Survey of Computer Vision-Based Human Motion Capture

Moeslund¹,

Granum²

2001

Computer Vision and Image Understanding

1,422

642

View full text Add to dashboard Cite

Section: Direct Model Usementioning

confidence: 99%

Section: Tracking Over Timementioning

confidence: 99%

A Survey of Computer Vision-Based Human Motion Capture

Moeslund¹,

Granum²

2001

Computer Vision and Image Understanding

1,422

642

View full text Add to dashboard Cite

“…Most human pose trackers rely on articulated kinematic models. Early generative models were specified manually (e.g., with joint limits and smoothness constraints), while many recent generative models have been learned from motion capture data of people performing specific actions (e.g., Choo and Fleet 2001;Herda et al 2005;Pavlović et al 1999;Sidenbladh et al 2000;Sminchisescu and Jepson 2004;Urtasun et al 2006;Wachter and Nagel 1999). Discriminative models also depend strongly on human motion capture data, based on which direct mappings from image measurements to human pose and motion are learned (Agarwal and Triggs 2006;Elgammal and Lee 2004;Rosales et al 2001;Shakhnarovich et al 2003;Sminchisescu et al 2007).…”

Section: Related Workmentioning

confidence: 99%

Physics-Based Person Tracking Using the Anthropomorphic Walker

2009

View full text Add to dashboard Cite

We introduce a physics-based model for 3D person tracking. Based on a biomechanical characterization of lower-body dynamics, the model captures important physical properties of bipedal locomotion such as balance and ground contact. The model generalizes naturally to variations in style due to changes in speed, step-length, and mass, and avoids common problems (such as footskate) that arise with existing trackers. The dynamics comprise a two degreeof-freedom representation of human locomotion with inelastic ground contact. A stochastic controller generates impulsive forces during the toe-off stage of walking, and springlike forces between the legs. A higher-dimensional kinematic body model is conditioned on the underlying dynamics. The combined model is used to track walking people in video, including examples with turning, occlusion, and varying gait. We also report quantitative monocular and binocular tracking results with the HumanEva dataset.

show abstract

“…Work on identification/filtering of hybrid systems first appeared in the seventies (see [19] for a review). More recent works consider variations of Problem 1 in which the model parameters, the discrete state and/or the switching mechanism are known, and concentrate on the analysis of the observability of the hybrid state [2], [4], [9], [11], [18], [21], [22] and the design of hybrid observers [1], [3], [7], [8], [10], [12], [14], [16], [17], [20].…”

Section: Problemmentioning

confidence: 99%

Identification of PWARX hybrid models with unknown and possibly different orders

Vidal

2004

Proceedings of the 2004 American Control Conference

View full text Add to dashboard Cite

Abstract-We consider the problem of identifying the orders and the model parameters of PWARX hybrid models from noiseless input/output data. We cast the identification problem in an algebraic geometric framework in which the number of discrete states corresponds to the degree of a multivariate polynomial p and the orders and the model parameters are encoded on the factors of p. We derive a rank constraint on the input/output data from which one can estimate the coefficients of p. Given p, we show that one can estimate the orders and the parameters of each ARX model from the derivatives of p at a collection of regressors that minimize a certain objective function. Our solution does not require previous knowledge about the orders of the ARX models (only an upper bound is needed), nor does it constraint the orders to be equal. Also the switching mechanism can be arbitrary, hence the switches need not be separated by a minimum dwell time. We illustrate our approach with an algebraic example of a switching circuit and with simulation results in the presence of noisy data.

show abstract

A dynamic Bayesian network approach to figure tracking using learned dynamic models

Cited by 176 publications

References 15 publications

A Survey of Computer Vision-Based Human Motion Capture

A Survey of Computer Vision-Based Human Motion Capture

Physics-Based Person Tracking Using the Anthropomorphic Walker

Identification of PWARX hybrid models with unknown and possibly different orders

Contact Info

Product

Resources

About