Introduction to Computer Vision Project Page (4005-757) - Fall 2007_1

NOTE:
1. All students must work individually
2. No more than 2 students can choose the same project. Once you decide on your project email your selection to rsg [at] cs.rit.edu
3. Students names will be posted on this webpage
4. Once two students choose a project the project will be listed as 'closed' and no longer available
5. Selection will be determined by date and time of email (see #2 above)
The remainder of this webpage documents other aspects of the project requirements

NOTE: A project is one component of the Introduction to Computer Vision, 4005-757 course

The computer vision project will give you the opportunity to apply what you have learned to an interesting and challenging problem. It is required that your project will extend the material presented in class. There should be a research component and a strong programming component. You must use the MATLAB programming language. The following table lists several project ideas. Choose a project from the list that  interests you.

• Proposal posted on your webpage October 4th before 10am. Also turn in a hardcopy in class including your project page URL..
  
• All project final write ups are due November 6th, 10am
• A printed MS Word document and a CD containing the report, all MATLAB source code, image databases and presentation in MS Power Point
  
• Graduate students must work individually. All sources used (papers, reports, Internet, etc.) must be properly credited and documented.
  
• Students enrolled in 4005-757 are required to present their projects during the last two weeks of class.  For the presentation should be 20 minute - 25 minutes.  These times are only estimates. Actual times will depend on the number of graduate students presenting which will be determined after the course withdrawal date.
• A schedule is (will be) posted on the course webpage.
  

Project Proposals (Due October 4th, 10am)
Project proposals are a ONE page document with the following sections:

Section 1	TITLE	A short descriptive title for the project.
Section 2	STUDENT	Your name (or names for a team)
Section 3	OVERVIEW	Briefly describe your project goals. From this overview it should be clear what you are going to do and the scope of the project.
Section 4	BACKGROUND	What sources will you use - papers, web sites, source code from sources other than what your team writes. A good source for finding articles: http://citeseer.ist.psu.edu/. Also use the library's databases, especially ACM, IEEE
Section 5	DATA	What data will you use. Where will it come from (course webpage, web? collect yourself?)
Section 6	WEB PAGE	URL of web site where you will record your progress - data, processed images, source code, etc. Your web page MUST be updated weekly. Weekly progress will be considered in determining your final project grade. The project will be graded out of 100 points. As indicated on the course calendar each webpage will be checked weekly for progress. Zero to 5 points will be assigned based on the progress reported. Progress includes - summary of papers read, description of the algorithm coded and test results. It is NOT adequate to state that you read a paper - you must summarize the key points of the paper and describe how it relates to your project. Provide a link to the paper if it is available on the web. Intermediate algorithm results must be reported weekly with image data results.

Final Write up Guidelines, Requirements, Code and Grading:



Use the same format that you find in professional journals, such as IEEE or ACM publications. At a minimum the paper should contain the following sections. Label each section.

• Your name and email address
  
• Abstract:  Clear summary of the work done and your results
• Section 1: Previous Work -  summarizes what has been done before in this area
  
• Section 2: Introduction - describes what you did and how it relates to the previous work
  
• Section 3: Algorithm - describes in detail what your approach was, what you actually did, your database, methods, algorithms
• Section 4: Results - carefully describes your results and compares it with previous published work in this area. Where did your work succeed, where did it fail?
• Section 5: References - cite all references. Plagiarism will result in a zero grade for the project. 
  
• Provide the link to your project web page
  
• The report should be in the 6-8 page range, 12 point type (do not fill all the pages with images- only use images as necessary)
• Plagiarism is a serious academic offense. You must write your report in your own words. You cannot work with another student on he project. If two students are doing the same project they cannot discuss their work with each other.  If you take an idea from another person you must reference that person. Use quotation marks were necessary (direct quotes). Violations of this policy will result in a grade of 0 for the project.
  

Project Grading - Activity	%
Complete Project Proposal posted on web and hardcopy turned in during class	10
Weekly Project Updates posted on Webpage (must show actual progress, see Section 6 above) Five points/week for 3 weeks as shown on course calendar	15
Difficulty of project, level of effort, success of project and functioning code	50
Quality of Final Report (Style, writing quality, references, etc.- see above)	25

Functioning Code

Project must be written in MATLAB

All code necessary to run your project must be included on the CD you turn in

Your main program must be called: mainYourName (insert your own name here) Your code must run directly from your CD. When I run mainYourName with the necessary parameters your code must run unassisted. The results must appear on the screen.

Project Ideas:

	Project Name	Description	Students
1	Super Resolution	See papers on super resolution by W.T. Freeman M. Irani and S. Peleg. 1991, "Super Resolution From Image Sequences" ICPR, 2:115--120, June 1990 S. Borman and R. Stevenson. Super-resolution from image sequences — a review. Technical report, University of Notre Dame, 1998. V. Cheung, B. J. Frey, and N. Jojic. "Video epitomes". In Proc. IEEE Conf. on Computer Vision and Pattern Recognition (CVPR), 2005.	1.  2.
2	Face Location and Feature Detection THIS PROJECT IS NOW CLOSED TO ADDITIONAL STUDENTS	Use a digital camera to create a database of images containing face. Or, collect images off the web but be sure there are no copyright issues. Develop an algorithm that will draw boxes around the faces in an image. For each face region, locate the eyes, nose and mouth regions. Mark and label each feature Do not use simple frontal facial images. The images should not be 'police style mug shots'	1. Fazim Mohammed  2. Rob Brambley
3	Man Made Object Detection THIS PROJECT IS NOW CLOSED TO ADDITIONAL STUDENTS	Develop an algorithm that will locate made made objects in an image. Straight lines or edges are a common features of man made objects. Segment the actual object and draw a box around it. A database is available on the course webpage	1.Bazarbekova Ainur  2. David Rubel
5	Recognition of Street and Road Signs in Images THIS PROJECT IS NOW CLOSED TO ADDITIONAL STUDENTS	Recognize simple geometric shapes in an image - rectangles, circles, etc. Apply algorithm to finding street signs, such as stop signs, speed limit signs, etc. Use your own images and images on course webpage	1.Vicky Steck  2. Amani Alsaqqaf
6	Segment  Objects Inside of a Refrigerator	Create a database of refrigerator (inside) pictures. Segment the different objects.  Attempt to identify objects (based on color, shape, etc.).	1.  2.
7	Moving People	Locate individuals walking in a video sequence. Draw a bounding box around the individuals	1.Almir Sehic  2.
8	Expert Object Recognition	Based on Baek's Ph.D. see: www.cs.colostate.edu/~vision/biomimetic/publications.html	1.  2.
9	Evaluate and Compare Interest Point Detectors	The Harris Corner Detector is a good starting point for this project. See the  following papers: "Saliency of Interest Points under Scale Changes" Daniela Hall, Bastian Leibe and Bernt Schiele "SUSAN corner detector", S M Smith and J M Brady, SUSAN a new approach to low level image processing, Intl. Journal of Computer Vision, 23(1):45 image processing, Intl. Journal of Computer Vision, 23(1):45-78, 78, May 1997 note: old paper, look for more recent publications	1.  2.
10	Image Registration	Determine mapping points between two images.	1.Bakytzhan Bitemirov  2.
11	Representation of Spatial Relations between Objects	Extract compact representation of spatial relations from images	1.  2.
12	Content Based Image Retrieval (CBIR)	Investigate, program and test techniques for image retrieval from a database	1.Haohui Yin  2.
13	Motion Detection in Video	Motion detection in video	1.Maxat Maketov  2.
14	Stereo Images THIS PROJECT IS NOW CLOSED TO ADDITIONAL STUDENTS	Estimate the distance to objects using a pair of stereo images. Explore different techniques	1. Dan Krisher  2. James Phipps
15	Image Resizing by Image Carving THIS PROJECT IS NOW CLOSED TO ADDITIONAL STUDENTS	.Shai Avidan, Ariel Shamir Seam Carving for Content-Aware Image Resizing ACM Transactions on Graphics, Volume 26, Number 3,   SIGGRAPH 2007	1.Andy Elble  2. Kurt Alfred Kluever
16	Video Surveillance	Track a moving object in a video sequence	1.  2.
17	Car Segmentation THIS PROJECT IS NOW CLOSED TO ADDITIONAL STUDENTS	Segment cars in 'street scene' images.	1.Karthikeyan Suryanarayanan  2. Mikkin Patel