Semantic analysis or annotation of videos is most useful when done frequently enough to capture the significant moments of a video, but not so frequently that annotations become busy and repetitive. With current techniques, semantic analysis is done too often, overloading the semantic analyzer and overwhelming the viewer with frequent, repetitive, or similar annotations of insubstantially differing frames. This disclosure presents techniques that detect substantial changes in the video for the purposes of semantic analysis. Timely and relevant annotations are presented to viewers without overwhelming them and without overloading the semantic analyzer.

Creative Commons License

Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.