A hybrid approach for summarization of cricket videos

This study proposes an automatic method for key-events detection and summarization for cricket videos, particularly because of the longest match durations, broadcasting time concerns and largely available multimedia content. In the proposed work, first rule-based induction is applied to detect excited audio clips in cricket videos, and then a decision tree framework is designed for video summarization. The proposed method evaluated on a diverse dataset with average accuracy of 95% signifies the effectiveness in terms of video summarization. Hence, the cricket videos can reliably be broadcasted over the low-bandwidth networks and transmission with time constraints.