Evaluation of Multimedia Content Summarization Algorithms

This paper provides an overview of 2 different methods for evaluation of multimedia (video) summarizations. A tag-based method focuses on tags selected for an analyzed video sequence. It checks if a summarized video sequence contains important content. An annotation method uses an annotation procedure to exact the key shots from full video sequences. Both methods used collectively allow for complex evaluation of summarizing algorithms. The paper contains descriptions and test results for both methods. The paper concludes with some suggestions for future directions.