Efficient Semantic Video Annotation by Object and Shot Re-Detection

3rd International Conference on Semantic and Digital Media Technologies, Koblenz, Germany

Manual video annotation on shot and on object level is a very time consuming and therefore cost intensive task. Automatic object and shot re-detection is one step forward in order to provide a cost efficient solution for temporally detailed video annotation. In this demonstration a tool will be shown which integrates novel video visualisation, navigation and interactive object re-detection techniques. Automatically re-detected objects and shots can be annotated by reusing operator annotations in a semi-automatic way. Object and shot re-detection capabilities are demonstrated on a variety of content, e.g. anchor person redetection in news content and setting and re-take detection in rushes content.

PDF