We present ICSI's research on large-scale video search methods using an application that reveals the geo-location of a consumer-produced video based on its content. We will show the demo web interface with three components: original video and its metadata, the types of visual and/or textual cues that were used to estimate the location of the video, the estimated location of video on the map.
As the fastest-growing type of content on the Internet, consumer-produced videos are a wealth of information about the world that's essentially untapped. The work I am presenting will allow research at never-before-seen scale for any empirical discipline.
NSF BIGDATA, NSF TC and NGA NURI Grant.