This document discusses various techniques for visual object tracking, including tracking whole objects, medium/fine level features, and facial feature points. It covers representation of tracked objects using templates, contours, and other models. Evaluation methods like normalized cross-correlation and boosted detectors are introduced. Simple tracking strategies like global search and contour tracking are described alongside their limitations. More advanced techniques like Lucas-Kanade tracking using gradient descent, mean-shift tracking, and regression-based tracking using linear and non-linear predictors are summarized. The use of motion models like the Kalman filter to incorporate temporal consistency is also mentioned.