Interesting section identification device, interesting section identification method, and interesting section identification program
AssigneePanasonic Intellectual Property Corporation of America
InventorKazuhiko Maeda
Life of this patent
Estimating the term…
Abstract
An interesting section identification device includes an object detection unit that detect an object included in a frame extracted from a video, a motion feature value extraction unit that calculates a motion feature value of the object, a distinctiveness calculation unit that calculates a degree of distinctiveness in terms of motion of the object with respect to the frame, and an interesting section identification unit that identifies an interesting section in the video with use of respective calculated degrees of distinctiveness of successive frames.
Claims20
An interesting section identification device that identifies an interesting section in a video that is estimated to interest a user, the interesting section identification device comprising:
- an object detection unit configured to detect an object included in a frame extracted from the video;
- motion feature value extraction unit configured to calculate a motion feature value of the object in the frame;
- a distinctiveness calculation unit configured to calculate a degree of distinctiveness in terms of motion of the object with respect to the frame, with use of the calculated motion feature value of the object in the frame;
- and
- an interesting section identification unit configured to identify the interesting section in the video, with use of the calculated degree of distinctiveness,
- wherein the object detection unit detects the object from each of a plurality of frames extracted from the video,
- the motion feature value extraction unit calculates a motion feature value of the object in each of the frames,
- with respect to each of the frames that is a target frame, the distinctiveness calculation unit calculates a degree of distinctiveness of the motion feature value of the object in the target frame from the respective motion feature values of the object in all other of the frames, and
- the distinctiveness calculation unit calculates the degree of distinctiveness of the motion feature value of the object in the target frame, based on an inner product of the motion feature value of the object in the target frame and an average of the respective motion feature values of the object in all other of the frames.
An interesting section identification device that identifies an interesting section in a video that is estimated to interest a user, the interesting section identification device comprising:
- an object detection unit configured to detect an object included in a frame extracted from the video;
- a motion feature value extraction unit configured to calculate a motion feature value of the object in the frame;
- a distinctiveness calculation unit configured to calculate a degree of distinctiveness in terms of motion of the object with respect to the frame, with use of the calculated motion feature value of the object in the frame;
- and
- an interesting section identification unit configured to identify the interesting section in the video, with use of the calculated degree of distinctiveness,
- wherein the distinctiveness calculation unit calculates the degree of distinctiveness of the motion feature value of the object from a motion feature value of other object included in the frame.
An interesting section identification method for use in an interesting section identification device that identifies an interesting section in a video that is estimated to interest a user, the interesting section identification method comprising:
- an object detecting step of detecting an object included in a frame extracted from the video;
- a motion feature value extracting step of calculating a motion feature value of the object in the frame;
- a distinctiveness calculating step of calculating a degree of distinctiveness in terms of motion of the object with respect to the frame, with use of the calculated motion feature value of the object in the frame;
- and
- an interesting section identifying step of identifying the interesting section in the video, with use of the calculated degree of distinctiveness,
- wherein the object detecting step detects the object from each of a plurality of frames extracted from the video,
- the motion feature value extracting step calculates a motion feature value of the object in each of the frames,
- with respect to each of the frames that is a target frame, the distinctiveness calculating step calculates a degree of distinctiveness of the motion feature value of the object in the target frame from the respective motion feature values of the object in all other of the frames, and
- the distinctiveness calculating step calculates the degree of distinctiveness of the motion feature value of the object in the target frame, based on an inner product of the motion feature value of the object in the target frame and an average of the respective motion feature values of the object in all other of the frames.
A non-transitory computer-readable recording medium storing an interesting section identification program for causing a computer to perform interesting section identification processing of identifying an interesting section in a video that is estimated to interest a user, the interesting section identification processing comprising:
- an object detecting step of detecting an object included in a frame extracted from the video;
- a motion feature value extracting step of calculating a motion feature value of the object in the frame;
- a distinctiveness calculating step of calculating a degree of distinctiveness in terms of motion of the object with respect to the frame, with use of the calculated motion feature value of the object in the frame;
- and
- an interesting section identifying step of identifying the interesting section in the video, with use of the calculated degree of distinctiveness,
- wherein the object detecting step detects the object from each of a plurality of frames extracted from the video,
- the motion feature value extracting step calculates a motion feature value of the object in each of the frames,
- with respect to each of the frames that is a target frame, the distinctiveness calculating step calculates a degree of distinctiveness of the motion feature value of the object in the target frame from the respective motion feature values of the object in all other of the frames, and
- the distinctiveness calculating step calculates the degree of distinctiveness of the motion feature value of the object in the target frame, based on an inner product of the motion feature value of the object in the target frame and an average of the respective motion feature values of the object in all other of the frames.
Dependencies are inferred from the claim text.