I came in today to try to finish all the parsing. It wasn't completely done on Friday, and that is necessary if we want the structure statistics to be displayed. I basically finished extracted all of the information from the files after some hours in the lab, but I doubt I will be able to compute all the necessary information and display it in html format by tomorrow's meeting. I am supposed to meet with Clifford tomorrow morning at 10:30 so that he might be able to write the html files while I work on the computation.

The parsing was completely done today, but extracting the info has proven to be a little bit difficult. I have noticed that the STL list class isn't working as I would expected (as in, it doesn't seem to really work), so I am hoping I do not have similar problems with the vector class that basically appears everywhere in my code.

Clifford missed meeting with me this morning, so I suggested that we cancel today's meeting since it didn't seem that we could really look at what we wanted to, and Clifford, Jenny, and I agreed to meet tomorrow morning to get everything done. In theory, this shouldn't be too difficult. But theory and implementation seem so far off from one another sometimes...

We met up this morning. Jenny agreed to make the links that I needed, and Clifford and I were going to write to the html pages. We didn't quite finish the html writing before the meeting that we had today. We promised to finish that up this afternoon, since it didn't seem as though it should take too long.

We had another meeting this morning because Sal is going away for the weekend and won't be back until Monday. We went through all the structure images' output, and found that the structure detector was pretty good at detecting buildings (homes) and fairly decent at not indicating structure in natural settings. Everyone is going to try to take pictures of office buildings, homes surrounded by trees/bushes, and skyscrapers with different skies. Clifford is going to work on organizing the pictures a little more, for instance putting all the townhouses in the same folder, all single family homes, etc.

The output to the html files that I was writing isn't quite right. It seems to think that all the lines in a particular cluster are of the same orientation...maybe I am not updated an iterator or something. Plus, I don't display them as a probabilistic density function, as we had discussed. Finally, I was using the area from the ellipse region, but apparently that is highly inaccurate and so I should be computing the actual area of the polygon. I have asked Yi if he could just include that in his XML output, since finding the area of a polygon given a set of points isn't trivial. It is doable, but not trivial.

Got the histograms to work! I was right about not incrementing an iterator. All better now. Yi seems to have promised to output the area of the polygon region for each cluster in the structure files, so I will just wait for him to do that, reparse parts of it, and rerun my code on all the various directories. I also made the output as a probabilistic density function now. I think I want to reorient the tables, though, so that the boxes of ranges are beneath, probabilities above, and alignment horizontal. I will do that when I return.

Went to Vancouver with roomates today!

Coordinated what needs to happen at this point with Jenny and Sal. Read more about multiclassification.