Mr. Burghouts will speak about recognizing human activities in real-time in thousands of videos
Video data are increasing: traditionally by camera networks and CCTV. Today, even smartphones have high-quality cameras. Cisco forecasts that by 2016, it will take 6 million years to watch one month of video data that crosses the internet. We need automated tools to categorize videos and to trigger us when they have a particular content of interest. One such tool is to search for particular human activities. Imagine how you are searching for a video of you playing basketball, or when your favorite soccer team scores a goal, or when your child was jumping on a trampoline. In 2010, the team of Gertjan Burghouts won a research grant in a DARPA program (2010-2013) about 48 human activities. Their technology performed best at multiple international benchmarks. In this talk, he will show many videos of their technology in action.
Gertjan Burghouts, PhD, received the MSc degree at the University of Twente (2002), and the PhD degree at the University of Amsterdam (2007) on the topic of recognizing objects in digital photos. In 2010, he won a $2.3M research contract at DARPA, in a research program about visual intelligence. In this program, his team developed a system to recognize 48 human behaviors in thousands of videos. His team performed best at the program’s benchmarks in 2011 and 2012. The system’s performance is state-of-the-art in various international benchmarks. His publishes in renowned journals and conferences and his work has been cited more than 700 times since 2005.