Abstract:

When we see things, our brains compute a description of the 3D external world from the 2D images received by the eyes. Although the problem has been solved by the brains of various kinds of animals and us human beings, it is still too difficult for modern computers. To understand visual computation, it is important to know what is the "equation" that vision is trying to solve, and what is the algorithm used for solving this "equation". In this talk, I will argue that seeing is statistical modeling and computational inference.