Semantic events detection and classification for baseball videos

This study presents an integrating decision approach, based on a model-based and rule-based decision approach, to detect and classify highlight events in baseball games. The model-based approach is proposed to model the relationship between the change statuses of the caption data and some events. The approach uses some binary bits to represent the amount of the changes. The rule-based approach is proposed to use the conditions of the caption data to elicit the empirical knowledge and to decide what some events is happened. Finally, the proposed approach integrates the complementary and redundant resulting of the two approaches to enhance the classification performance. Experimental results demonstrate that the proposed approach is very efficient for classifying highlight events. Furthermore, the performance of the proposed approach can be improved by 6.58% over the average of the model-based and rule-based approaches.