Week 3 Log

• We gave Project presentation 1 on 13/06/2016. It can be found here.
• Created Tutorial on Object Tracking (Based on ROI) using CAMShift algorithm with 2D histogram. It can be found here.
• As in CAMShift we can not re-track the object if it escapes and comes back into the frame so we searched for other object tracking algorithms.

Matrioska Tracker

• We got Matrioska tracking algorithm. Which uses keypoints (feature) matching and learning to track the object. Its details can be found here
• We tried to understand Features, Descriptors, Feature Matching. Some information can be found here
• We created code for tracking using SIFT, SURF feature matching and homography.
• Tested this code on some objects and got problems like "It is very slow", "If object moves it don't track object", "It gives bounding box only when object is static", "SIFT and SURF are patented algorithms".
• After gathering required information, we tried to implement Matrioska tracker and created a code that can be found here
• Tested this code on various objects, got problems as "It is not filtering background keypoints" etc.