Exploring the Synergy of Humans and Machines in Extreme Video Retrieval

We introduce an interface for efficient video search that exploits the human ability to quickly scan visual content, after automatic retrieval has arrange the images in expected order of relevance. While extreme video retrieval is taxing to the human, it is also extremely effective. Two variants of extreme retrieval are demonstrated, 1) RSVP which automatically pages through images with user-control of the paging speed, while the user marks relevant shots and 2) MBRP where the user manually controls paging and adjusts the number of images per page, depending on the density of relevant shots. 1 Interactive vs. Automatic Video Search When comparing results of fully automated video retrieval to interactive video retrieval [5], one finds a big gap in performance. The fully automated search (no user in the loop) succeeds with good recall for many topics, but relevant shots tend to be distributed throughout the top 3000 to 5000 slots in the ordered shot list, causing the standard metric of a...

• Type: Conference
• Year: 2006
• Where: CIVR
• Authors: Alexander G. Hauptmann, Wei-Hao Lin, Rong Yan, Jun Yang 0003, Robert V. Baron, Ming-yu Chen, Sean Gilroy, Michael D. Gordon
• Keywords: CIVR 2006, Extreme Video Retrieval, Image Analysis, Relevant Shots, Video Retrieval