Automatic Soccer Video Key Event Detection and Summarization Based On Hybrid Approach

Use of hybrid approach to summarize and detect key events of soccer

Abstract

Sports broadcasters generate an enormous amount of video content in cyberspace due to massive viewership all over the world. Analysis and consumption of this huge repository urge the broadcasters to apply video summarization to extract the exciting segments from the entire video to capture the user's interest and reap the storage and transmission benefits. Therefore, in this study, an automatic method for key-events detection and summarization is presented for soccer videos. The proposed framework reaps benefit both from learning and non-learning methods of summarization. SVM classifier is used to classify boundary and non-boundary frames based on extracted features .Histogram difference and the average motion vector. Replay detection shot view classification, and play break sequence formation are performed through efficient algorithms. Nonsubjective features such as play break duration ratio, play duration ratio, near goal duration ratio, etc. are used to perform statistical analysis which helps in devising rules for summarization. To address the shortcomings associated with key event detection and summarization algorithms and to get the best out of the merger of learning and non-learning based approaches for summarization, this research problem needed a deep inside look. The effectiveness and robustness of the proposed method are tested over an extensively
huge dataset and the results are highly productive and comparative. Besides soccer, the proposed method finds its application in freestyle football and hockey but with minor modifications.