Gaze location prediction in broadcast football video
Cheng Qin, Dimitris K. Agrafiotis, Alin Achim, D.R. Bull · 2012
In this work we investigate gaze location prediction for the context of broadcast football video. We use Bayesian integration to combine bottom up (color, intensity, orientation, motion, novelty) and top down (ball) features. 70% success rate is obtained for CIF.