5. Conclusion While there are numerous other means to visualizing the data in class videos, CVNView focused mainly on three implementations: a Slide View, an Information Mural, and a Box View. These three views provide a Zoom - Filter - Details-on-demand method of exploring the data. The Box View focuses on the larger data structure, without giving details about the slides, by displaying sections and topics as boxes in a two-dimensional grid. The Information Mural then zooms in on the Box View by showing some more detail about each section and the therein contained slides. The Slide View then lays out the image data of all slides as thumbnail images, and, finally, also allows the user to use a virtual magnifying glass to browse the slides. For this version of CVNView, the slide data has been manually prepared for two classes. This preparation is time consuming and very subjective in the segmentation of data, as discussed earlier. In order for this or any similar application to be used on a larger scale, the process of slide data segmentation must be automated. This requires an analysis of the image data for each slide to the extent of recognizing how slides can be grouped according to topic. While the human eye can recognize similarities between slides by interpreting the slide data, an image analysis may have to focus on interpreting color information in the slide, yet the color information changes constantly given that the content of a Board or a sheet of paper changes as the topic progresses. External factors, e.g. lighting of the room or poor image quality, may undermine this method of image analysis. Given the different media types, image data must also be interpreted in different ways. The task of automated segmentation is not only subjective in analyzing image data, but must also be customized for the purpose, as in the case of CVNView for the environment of the Columbia Video Network.