Segmentation and Fitting using Probabilistic Methods
Missing data problems, fitting and segmentation, including the EM algorithm
The EM algorithm in practice, including image segmentation, line fitting, motion segmentation, identifying outliers, background subtraction, the fundamental matrix and difficulties with the algorithm
Model selection, including AIC, BIC, description length and other methods for estimating deviance
Tracking as an abstract inference problem, including independence assumptions, tracking as inference, and an overview
Linear dynamic models, including drifting points, constant velocity, constant acceleration, periodic motion and higher order models
Kalman filtering, including the Kalman filter for a 1D state vector, the Kalman update equations for a general state vector and forward backward smoothing
Data association, including nearest neighbours, gating and probabilistic data association
Applications and examples, including vehicle tracking
