2.2 Data Acquisition
CVN provides one directory per class, which contains all slides as JPEG images. All slides, with the exception of PowerPoint slides and the title slide, are numbered consecutively 1 through n, i.e. [1.jpg, 2.jpg, ...] . The title slide is named title.jpg, and PowerPoint slides are named hoy.ppt.slideX (X = [1..m]). PowerPoint slides may appear at any point in the sequence of slides without breaking the numeric sequence, e.g. {67.jpg 68.jpg hoy.ppt.slide6.jpg hoy.ppt.slide7.jpg 69.jpg}. It is thus impossible to deduce the correct sequence of slides by merely looking at the slide names for a given class that contains PowerPoint slides, without using some kind of index. For the purpose of ordering the slide data, we make use of the already existing interface from CVN for viewing the class material:
|
 |
| Figure 2: CVN Interface: Index of Key Frames |
|
2.3 Segmentation of Data
In order to represent all of the slide data in some meaningful visual context, the slides must first be placed into categories. For the first level of classification, we simply divide the slides by media type: Board, Class, Computer, Podium, and Sheet, and place each slide into one of 5 directories to this effect. Within these directories we then divide the slides by topic: we analyze each slide, and decide on whether a given slide falls into the category of another slide given context and/or location. Depending on the types of media used, this second classification can be very subjective and is thus far done entirely manually (see discussion on each of the 5 media types below).
The slides need not be separated further into sections. (see "Sections: Breaks in the Sequence of a Topic" below)
|
