Personalized Sports Video Customization Based on Multi-modal Analysis for Mobile Devices

In this paper we have implemented a mobile personalized sports video customization system for mobile users by using a novel approach. The system is based on the B/S architecture, with this architecture, the whole system can be divided into two parts: a friendly client browser on smart phones and an intelligent multimedia analysis server. For the client browser, a friend UI is designed for mobile users to customize their favorite video clips. For the multimedia server, with the analysis of video content as well as user preference, semantics-specific video clips can be tailored to the particular users to effectively improve their visual enjoyment. The whole approach of the multimedia analysis is implemented in a two steps. First, sports video content is automatically indexed by detailed textual descriptions, which are derived from latent semantic analysis of web-casting text, based on the semantic annotation, a video database is constructed. And then video content and user preference are collaborated into event importance computation mechanism, after that the server can tailor the optimal set of video clips to the particular user. Particularly, we adopt basketball and football games as our initial sports genres because they not only widely adopted study bed but also globally popular sports, which possess great values in both research and application. Both quantitative and qualitative experiments validate the effectiveness of our system.