Research Papers

Throughout the course of the semester we will be reading research papers that pertain to the topics we are covering in lecture. You will be responsible for reading each paper and writing a mock-review of the paper. You will also be responsible for presenting (with a partner) one paper over the course of the semester.

Paper Schedule
Review Details
Presentation Details

Paper Schedule

Date	Topic	Paper	Presenters
Mon, Sept 19	Segmentation	Contour and Texture Analysis for Image Segmentation by Malik, Belongie, Leung, and Shi [MBLS01]	Ben and Ricco
Mon, Oct 3	Object Recognition	Eigenfaces for Recognition by Turk and Pentland [TP91]	Rebecca
Wed, Oct 19	Context in Recognition	Contextual Priming for Object Detection by A. Torralba [T03]	Kevin and David
Mon, Oct 31	Sketch/diagram Recognition	Combining Geometry and Domain Knowledge to Interpret Hand-Drawn Diagrams by Gennari, Kara, Stahovich and Shimada [GKSS05]	Andrew and Chris
Mon, Nov 14	Applications	Utilizing Segmented MRI Data in Image-Guided Surgery by Grimson et al. [GEKLWK97]	Dan and Mike

Bibliography

[MBLS01] Jitendra Malik, Serge Belongie, Thomas Leung and Jianbo Shi. Contour and Texture Analysis for Image Segmentation. International Journal of Computer Vision 43(1). pp. 7-27.

[TP91] Matthew Turk and Alex Pentland. "Eigenfaces for Recognition." Journal of Cognitive Neuroscience, 3 (1). 1991.

[T03] Antonio Torralba. "Contextual Priming for Object Detection". International Journal of Computer Vision, 53 (2): 153-167 2003.

[GKSS05] Leslie Gennari, Levent Burak Kara, Thomas F. Stahovich, and Kenji Shimada. Combining geometry and domain knowledge to interpret hand-drawn diagrams. Computers and Graphics 29 (4) 2005.

[GEKLWK97] E. Grimson, G. Ettinger, T. Kapur, M. Leventon, W. Wells, R. Kikinis, "Utilizing Segmented MRI Data in Image-Guided Surgery", Int. J. Pattern Recognition & Artificial Intelligence, special issue on Processing, Analysis, and Understanding of Magnetic Resonance Images of the Human Brain, 11(8):1367-1397, 1997

Details on the Review

For each paper EXCEPT for the one that you present, you should write a review of the assigned paper. The review should be about 1-2 pages in length. The format you should use is a short-answer rather than essay format. That is, you should specifically address each of the points below. In your writeup you should list the point, and then give a concise, informative response to the point. Your response may be a paragraph or two, or a bulleted list, whichever you feel helps you address the point best.

Here are the points you must address in your review:

Give a brief summary of the work.
What is this work's main contributions to the field of computer vision?
Is this work technically sound?
Is the writing clear?
How well do the authors position the work in the field of computer vision?
What (and how much) impact will this work have (has it had) on the field of CV? (Note: Because these papers have already been published, they (hopefully!) have already had an impact on the field of CV. If you can notice their impact in what you know about CV, state it. Otherwise, try to "predict" what that impact was/might be.)

To help you write this review, I give some general and specific tips below:

Keep your comments brief and to the point.
Back up your claims by citing specific examples from the paper whenever possible.
Write a rough draft, then go back and revise it just as you would any essay or paper.
Imagine that your comments will be read by the authors of the paper. Try to make your comments as helpful to them as possible.

Reviews are due before class on the day the paper is presented. Please submit your review electronically by emailing me a PDF (only!). If this is a problem for you let me know ASAP.

Presentation

When it is your turn to present, you should prepare your presentation to contain the following:

Summary of the paper, including information about where and when it was published, and who the authors are, and the major contributions it makes to its subfield of CV. This will be about the first 5 minutes of your presentation.
Presentation of the technical details of the paper. You should think of this part as "getting everyone on the same page". You should assume that everyone has read the paper, but that they might not have the same deep understanding that you do, because you have read it more carefully than they have in preparation to explain it to the class. ;) You will spend the majority of your time on this part of the presentation.
Discussion of the strengths/weaknesses, novelties, applications, etc. of this work. This part of the presentation is more free-form. The idea is to get an interesting discussion started about the ideas the authors present in the paper.

Your presentations should be about 45 minutes in length. You should both speak for some significant portion of this time. If they are slightly longer, that's not a problem, but try to keep them under an hour. This may sound like a lot, but the goal is to get others involved so that it's not just you up there talking the whole time.

Once again, here are some tips for success:

Start reading the paper early and come see me with questions on technical details.
Prepare your presentations early. You must come see me on Thursday or Friday before your scheduled presentation to go over your current status. Doing so will greatly improve the quality of the presentation.
Practice at LEAST once, preferably in front of someone else (but at the very least in front of each other).
Involve the class. Ask questions. Ask for opinions. Just get people talking.
If you get lost/confused/distracted while you are speaking, simply pause, acknowledge that you need a second (it's totally OK to say something like "Excuse me, let me back up for a second to get back on track". It happens to everyone.), and then regroup. The most important thing is to remain calm.