Skip to Main Content
With the advent of broadband networking, video will be available online as well as through traditional distribution channels. The merging of entertainment and information media makes video content classification and retrieval a necessary tool. To provide fast retrieval, content management systems must discern between categories of video. Automatic multimedia analysis techniques for deriving high-level descriptions and annotations have experienced a tremendous surge in interest. Academia and industry have also been challenged to develop realistic applications-from home media library organizers and multimedia lecture archives to broadcast TV content navigators and video-on-demand-in pursuit of the killer application. Current content classification technologies have undoubtedly emerged from traditional image processing and computer vision, audio analysis and processing, and information retrieval. Although terminology varies, the algorithms generally fall into three categories: tangible detectors, high-level abstractors, and latent or intangible descriptors. This paper presents the reflections of the work done by the author and the work ahead.