DEVICE AND METHOD FOR GENERATING LABEL OBJECT VIDEO OF VIDEO SEQUENCE

PROBLEM TO BE SOLVED: To provide a method/device for generating the label object video of a video sequence by which a query object area is extracts in respective frames by a key frame without regard to the movement of an object without requiring additional operation and a labeled object video is generated in the respective frames from corresponding query object information. SOLUTION: The video sequence including a query object is inputted (200, 210). The video sequence is divided into shots consisting of frames having similar scene characteristics (220). Whether the query object exists in a representative video expressing the respective shots is judged to extract the initial object area of the corresponding shot (230). By using these initial object areas, the object area is tracked and displayed to the frames existing in the respective corresponding shots (240) to generate the label object video (250).