Project1: Spatial Pyramid Matching for Scene Classification (Using bag-of-words (BoW) approach for scene classification)
Project2: Feature Descriptors, Homographies & RANSAC (Implementing an interest point (keypoint) detector (SIFT), a feature descriptor (BRIEF) and an image stitching tool based on feature matching and homography)
Project3: Lucas-Kanade Tracking (Implementing Lucas-Kanade (LK) tracker with one single template, generalizing the tracker to accommodate for large appearance variance (Lucas-Kanade Tracking with Appearance Basis). Implementing a motion subtraction method for tracking moving pixels in a scene (Affine Motion Subtraction). Efficient Tracking using Inverse Composition)
Project4: 3D reconstruction.
Project5: Neural Networks for Recognition