Graphical models provide an intuitively appealing interface for modeling highly-interactive sets 
of random variables. They lend themselves naturally to the design of efficient general-purpose 
algorithms (e.g. decoding Turbo codes). Due to their appealing properties, graphical models have 
become a popular tool for solving problems in various scientific areas including computer vision. 
In order to have a better insight in the design and inference of graphical models, we will study 
several cases, such as HMMs, quad-trees, and dynamic-structure trees, and discuss their performance 
in image classification.