I had a meeting today at the HIT (Human Interface Technology) Lab, which is on the second floor of Fluke Hall, with Sal, Jenny, Clifford, Linda, and me. We discussed the content-based image retrieval project generally, got somewhat updated on the current status of the project and prior research in this area, and outlined taks for this week more detailed outline can be found in the notes from today. (*Note: On the way to the meeting, Linda and I talked briefly about my future plans. I explained some of my interests, and she suggested that I email Raj Rao, an assistant professor here in the computer science department. *)

My contribution to this project will be setting up a multiple class structure in C++ that we will then use to subdivide areas of each image into different classes and hopefully use clustering techniques to create what we are calling an abstract image. We currently have structural, textural, and color images from which we would like to carve out classes. We are hoping that through clustering overlapping regions, and perhaps some more algorithmic complexity, we will be able to classify and label objects to yield a new image, or abstract image.

To do:

1. Get conference CD from Sal.
2. Refresh knowledge on probability.
3. Read award-winning Fergus et al. conference paper.

Today I have been doing more reading up on background for the project. I am also learning html to make these webpages, and thus keep all notes organized. I arranged to get an extra hard disk, since Yi's computer is nearly filled up. I think that will prove useful, even though it is a somewhat noisy hard drive. I also got the CD from Sal and we met together briefly, which was quite useful. He helped me to prioritize my background research.

After having looked at the probability chapters Sal sent out, which were compiled by Marina Meila, I have concluded that I remember the basics of probability but need to further review the concepts from handout 5---in particular, the maximum likelihood estimation. (*It is important to note here that Sal mentioned that I should meet with Marina at some point if I am interested in doing a PhD in computer science. She is in the statistics department here, but having looked at her webpage reveals that she has worked in the AI lab at MIT and did her post-doc at CMU. She can be contacted at: mmp@cs.washington.edu.*) Got an old probability book from Sal, but might want to also use Introduction to Probability.

To do:

1. Put together mini-presentation on Chapters 3 and 4.1 from probability handouts.
2. Reread Fergus article.
3. Get more literature from Linda.
4. Look up articles that were referenced in Fergus paper (see IEEE online).
5. Continue with yesterday's readings.

Today was the day that I actually began compiling these notes.

I got a folder of articles from one of Linda's classes today to review. I also went through the part of the probability handouts that I need to present. I know basically what I am going to say. I made a powerpoint presentation for my part.

Reread Fergus article. Makes a little more sense. Need to still look up the references from that article. Reread the information on maximum likelihood, but think I should use another, more user-friendly source to understand that. Or just wait until Friday's meeting.

To do:

1. Get references from the article.
2. Print out presentation and prepare.
3. Skim through Computer Vision book.
4. Look up maximum likelihood in a new source.

Looked up more information on maximum likelihood. Prepared, now, for tomorrow's presentation. Got the articles referenced in Fergus' paper from online -- should ask Sal which papers to focus on.

Got a big source on content-based image retrieval that I should look through (thanks Arati!). Got more done on the website that I need to make. Have figured out more html. Might want to start reviewing C++ soon, or at least find good online documentation, since Sal said we would probably start coding next week. To do:

1. Ask Sal which papers to focus on, from those found.
2. Ask about how to make transparencies for the future.
3. Ask about where the website will be located (both mine and the groups).
4. Find out about a mailing list for the group -- organize if possible.
5. Ask Linda about the brain group's meetings -- they sounded interesting, from what Arati said.
6. Find good online C++ documentation (or ask for books from Sal/Linda).
7. Ask Linda about more sources that I should be reading from her office.

Had a meeting with the 2D and 3D groups. Met Viet, who is working on the 3D project. Made more concrete plans for the 2D project that can be viewed in the notes from today.

Talked with Jenny after the meeting about her part. She agreed to update the website again. I suggested some changes to her. Also, for now, she is going to try to learn C++ (code up some little examples) while I look into the XML/C++ parser.

Found some C++ resources at Duke, the C++ resources network, and might be able to find something in free online C++ documentation.

To do:

1. Learn more about XML and maybe another XML/C++ parser option.
2. Get XML book from Sal.
3. Figure out the different labels in each different kind of image.
4. Design data structure for C++ (or at least get an idea of the layout).