Recognizing action at a distance-英文文献.pdfVIP

Recognizing Action at a Distance Alexei A. Efros, Alexander C. Berg, Greg Mori, Jitendra Malik Computer Science Division, UC Berkeley Berkeley, CA 94720, USA /˜efros/research/action/ Abstract Our goal is to recognize human actions at a distance, at resolutions where a whole person may be, say, 30 pix- els tall. We introduce a novel motion descriptor based on optical flow measurements in a spatio-temporal volume for each stabilized human figure, and an associated similarity measure to be used in a nearest-neighbor framework. Mak- ing use of noisy optical flow measurements is the key chal- lenge, which is addressed by treating optical flow not as precise pixel displacements, but rather as a spatial pattern of noisy measurements which are carefully smoothed and aggregated to form our spatio-temporal motion descriptor. To classify the action being performed by a human figure in a query sequence, we retrieve nearest neighbor(s) from a Figure 1. A typical frame from the NTSC World Cup broad- database of stored, annotated video sequences. We can also cast video that we use as our data. Humans are extremely good use these retrieved exemplars to transfer 2D/3D skeletons at recognizing the actions of the football players, despite the low onto the figures in the query sequence, as well as two forms resolution (each figure is about 30 pixels tall; see the zoomed in of data-based action synthesis “Do as I Do” and “Do as I player at the lower left corner). Say”. Results are