me

Intelligent Multimedia Systems 2010-2011

On this site you'll will find the latest information on the lab session of the course Intelligent Multimedia Systems.

The set up of these lab sessions is as follows. Every week, an assignment is made available on this web page. This purpose of this assignment is to help you reach the final goal of this course, which is to create your own tracking application in Matlab. These assignments are there to help you familiarize yourself with Matlab and with translating formulas and algorithms from papers to working Matlab functions. Translating formulas and algorithms from papers to a practical understanding is an important research skill, as is writing scientific papers. During the course, some hints and tips are made available on this website. Your deliverables for this course only includes a final scientific paper on the tracker (accompanied by the source code of your application and some videos with results of your tracker), so no report is required for the weekly assignments. See also the assignment page for a little bit of extra info on the paper.

Lab Schedule

All lab rooms are situated in Science Park 904.

Date Time Week Room Remarks
Monday September 13 13:00-15:00 37 B1.24FGHIJ
Monday September 20 13:00-15:00 38 B1.24FGHIJ Try to get a spot in room I or J so we're all together
Monday September 27 13:00-15:00 39 B1.24FGHIJ Lab 3
Monday October 4 13:00-15:00 40 B1.24FGHIJ Lab 4
Monday October 11 13:00-15:00 41 B1.24FGHIJ Lab 5
Monday October 18 13:00-15:00 42 B1.24FGHIJ Lab 6
43 No lab course (exam week for some courses)
Friday November 5 09:00-11:00 44 G0.10-G0.12
Friday November 12 09:00-11:00 45 G0.10-G0.12
Friday November 19 09:00-11:00 46 G0.10-G0.12
Friday November 26 09:00-11:00 47 G0.10-G0.12
Friday December 3 09:00-11:00 48 G0.10-G0.12
Friday December 10 09:00-11:00 49 G0.10-G0.12
Friday December 17 09:00-11:00 50 G0.10-G0.12
Friday December 31 23:59 52 Paper deadline
Note: I will answer questions by e-mail until December 24th

Lab Overview

During the course of this lab you will develop a Mean-Shift Tracker . The Mean-Shift approach to tracking is a state-of-the-art algorithm for tracking objects in videos. Since this is a quite a complicated algorithm, this lab will try to guide you through the development process.

The work you do in the lab sessions is not graded, only the work you hand in for the programming assignment is. The purpose of the lab sessions is to introduce you to the concepts you need to develop the tracking application. Each week new concepts and Matlab implementations are introduced. At the end of the course you should have enough experience to be able to implement the Mean-Shift Tracking application.

The Mean-Shift algorithm is described in Kernel-Based Object Tracking by Dorin Comaniciu, Visvanathan Ramesh and Peter Meer. Another paper, by the same authors on the same subject, is included in the Reader that accompanies this course. At this time the lab will cover the following subjects:

Part 1 Introduction to Matlab
Part 2 Histogram representation
Part 3 Object localisation with morphology/histogram backprojection
Part 4 Brute force object tracking
Part 5 TBD (Object classification with bag-of-words?)
Part 6 Kernel based tracking
Part 7 Mean-shift tracking
Intermezzo (TBC) Interactive Multimedia Search
Part 8 - 12 Finishing and evaluating the tracker