Goal detection from unsupervised video surveillance

Abstract

Unsupervised video surveillance that can automatically learn, predict or detect events can be useful in unsupervised video surveillance that can automatically learn, predict or detect events can be useful in many practical situations. This work describes how many practical situations. This work describes how an unsupervised surveillance can be used in goal detection in basketball videos. We present a system which takes as input a video stream of a basket and an agent trying to hit a goal and produce an analysis of the behavior of the ball in the scene and detect goals. To achieve this functionality, our system relies on two modular blocks. The first-one detects and tracks moving balls in the sequence. The second module takes as input these trajectories and makes decision on a goal versus non goal. We present details of the system, together with results on a number of real video sequences and also provide a quantitative analysis of the results. The approach described here uses object detection and mean-shift tracking to detect and track the basketball in a video. Goal decision is based on the positions of the ball, its current and immediate past positions, in image frame, with respect to a matrix representing the basket. © 2011 Springer-Verlag.