Blog for work on my Masters thesis - a survey of methods for evaluating media understanding, object detection, and pattern matching algorithms. Mostly, it is related to ViPER, the Video Performance Evaluation Resource. If you find a good reference, or would like to comment, e-mail viper at cfar.umd.edu.
Archives
Media Processing Evaluation Weblog
Thursday, April 24, 2003
Writing a new AppLoader
Man, I need to get my ADHD under control. To get ViPER v4 to support loading modules, the application loader must handle dependencies and dynamic menu systems elegantly. Both of these things are graphs; a partial order is a DAG and the menus are trees. So, instead of using something like BeanBox or adapting the NetBeans application toolkit, I wrote an application loader that uses RDF, the W3C's standard for representing semantic graphs. Now I have an RDF specification for a Java application, which specifies what beans to load and a partial order on how to load them, and a tree for loading file menu items. I haven't written any OWL for them yet; it is pretty ad hoc. I'll probably go through and clean up the ontology after I get it loading the Chronicle and Canvas views.
- posted by David @ 10:41 PM
Wednesday, April 23, 2003
Jonathan used IBM's Flavor to develop his MPEG-1 parser. Flavor is designed for streaming, and there were some bugs in the output, so he did modify the output considerably. Flavor is a very useful language. Fortunately, it is released under a variant of the artistic license, so we can use it as a module.
- posted by David @ 10:48 AM
Monday, April 14, 2003
I haven't really been meeting with Dave these last few weeks, as he's been busy with SDIUT 2003. This week, I hope to discuss activity detection evaluation.
- posted by David @ 2:43 PM
