Demo: Spatio-temporal template matching for ball detection

Abstract

This paper considers the detection of ball in a basketball game covered by multiple loosely synchronized cameras. First, plausible ball candidates are detected on the nodes of a 3D grid defined around the basket. This is done by correlating independently in each view the spatial template of the ball with a precomputed foreground mask. Efficient implementation of this step relies on integral images. Afterwards, false positives are filtered out based on a temporal analysis of the ball trajectory. This analysis builds on the Random Sample Consensus (RANSAC) method, with a ballistic trajectory model. The integrated approach is demonstrated on a real-life dataset, and appears to be both effective and efficient.