Unsupervised activity discovery and characterization from event-streams
Roszilah Hamid, Siddhartha Maddi, Amos Y. Johnson, Aaron Bobick, Irfan A. Essa, Charles Lee Isbell · 2005
We present a framework to discover and characterize different classes of everyday activities from event-streams. We begin by representing activities as bags of event n-grams. This allows us to analyze the global structural information of activities, using their local event statistics.