Skip to Main Content
As digital sports video data become more and more pervasive, automatic highlight detection is particularly demanding. We present an automatic and effective framework for goal event detection based on the cinematic features. The proposed framework includes 4 parts: shot boundary detection, shot classification, slow-motion detection, and goal event detection. The shot boundary detector employs the multi-filter structure. Then, the goal detector uses the slow-motion replay to locate the goal shot and the new cinematic template to model and identify the goal event. By the experiments over diverse soccer game videos, the proposed framework is effective and efficient to find the goal shot.
Machine Learning and Cybernetics, 2004. Proceedings of 2004 International Conference on (Volume:6 )
Date of Conference: 26-29 Aug. 2004