The main themes of the Fellowship are those of developing statistical models of the appearance of objects in images, and exploring the corresponding search techniques to allow new objects to be located in new images.
The original plan during this third year was to work on dealing with occlusion, dealing with multi-scale models and developing different sorts of flexible models. However, at the end of 1997 we created the Active Appearance Model. Most of 1998 has been spent understanding, developing and testing this powerful new algorithm.
In this report I have taken the advice of a reviewer of my 1997 Progress Report and used more than the recommended 500 words in order to give a more detailed report of my activities.