Using existing popular video services, a user must download and manually parse entire video files even if he or she is only interested in one particular scene or event within those videos. This approach wastes both bandwidth and time. An interesting problem is thus how to efficiently automate the video search process across a large video archive collection.
We have built several versions of Davvi, end-to-end prototypes of an Internet based video system that allows users to automatically search for and select specific events from across large video archives and combine them into a single logically composed video for almost immediate playback. Davvi does more than just returning the most relevant results obtained from an inverted index, it actually extracts video events from across a huge archive of larger videos and composes them into a smooth play-out. The application scenario is soccer, and our conjectures for Davvi grew out of close collaboration with our (former) industrial partner Schibsted.
Our industrial partner Microsoft has also been key for development of a video-enabled enterprise search platform (vESP). This is an application prototype that enhances the recently released enterprise search engine from FAST with video streaming. The idea is that in a large enterprise, like Microsoft, there exists a lot of information in form of presentations with corresponding video. Using our enhancements, a user can select and combine slides from different presentations generating a new slide deck dynamically, and the corresponding video clips are concatenated and presented on-the-fly aligned with the slides.