A Video Retrieval System Framework Based on SVM
基于SVM的视频检索系统框架①

Transformation from feature to semantic info is the important step of Video Retrieval. Based on the improved SVM procedure for Video Retrieval according to relevance feedback, reranking and BOOST, this paper proposes a framework. Using the data of TREC2007, the experimental result shows the framework has sufficient improvement in AP and MAP.