The course will focus on a number of computer vision technologoies including: camera geometry, object detection and recognition, and object tracking in video. In each case we will discuss the potential applications and the underlying technology that makes it possible. In particular, we will study the geometry of cameras (projective geometry, the fundamental matrix and bundle adjustment). Stereo and wide-base-line stereo (correspondence problem, SIFT features). Object detection and recognition (including the Bag-of-words model for object recognition, the Viola-Jones method for face detection and the Dalal-Triggs method for human detection). In addition, we will discuss methods for fast approximate nearest neighbors (LSH,kd-trees), quantization (k-means and mean-shift) and learning (AdaBoost). Finally, we will learn about object tracking in video (including mean-shift tracking, kalman and particle filtering).
Grade is based on:
Homework assignments & Matlab projects (50%)