Back to results
US 8,789,120 B2Granted
Temporal video tagging and distribution
AssigneeSony Corporation
InventorsSean Kennedy, Edward Winter
Life of this patent
Estimating the term…
Abstract
A method and apparatus for temporal tagging of videos in response to identification of locations on the video where persons in a contact list of a user are identified in response to facial recognition. The temporal tags can be retained within the video, or contained separately in a metadata file having fields to indicate time locations in the video and identified persons seen at those locations. Use of the temporal tags allows for automatic distribution of content relevant to the recipient, based on their presence in the video, or of persons in the video of which they are interested.
Claims20
An apparatus for processing a video, comprising:
- a computer;
- programming executable on said computer for processing a video received by said computer, said processing comprising:
- performing facial recognition of images in said video to identify one or more persons in said images;
- performing voice recognition of persons speaking within said video;
- comparing persons identified by said facial recognition and said voice recognition with persons contained in a contact library;
- generating temporal tags corresponding to locations in said video at which persons contained in said contact library are identified in said video using a combination of said facial recognition and said voice recognition;
- and
- storing said temporal tags in association with said video;
- wherein said temporal tags determine which portions of said video a recipient is able to subsequently view based on a correspondence between said temporal tags and identification of the recipient;
- wherein said temporal tags are utilized in combination with an overlap threshold and a persistence value when determining which portions of said video a recipient is able to subsequently view;
- wherein said persistence value establishes length of video played for the recipient after persons identified by facial or voice recognition are out of a scene;
- and
- wherein said overlap threshold is configured to increase persistence by making video sections contiguous which have close temporal proximity and the same identified persons;
- wherein an appearance reset threshold is configured with a longer time duration than said persistence value, in which said appearance reset threshold establishes length of time an identified person must be absent from a video segment before considering this position in the video as the end of their presence.
11 more elements
An apparatus for processing a video, comprising:
- a computer;
- programming executable on said computer for processing a video received by said computer, said processing comprising:
- performing facial recognition of images in said video to identify one or more persons in said images;
- performing voice recognition of persons speaking within said video;
- generating temporal tags for locations in said video at which a said person is contained in a separate contact library is identified in response to a combination of facial recognition and said voice recognition;
- wherein said temporal tags comprise indices that indicate locations within said video at which said person appears in said video, or are heard therein;
- storing said temporal tags within said video, or in a file associated with said video;
- and
- selecting portions of said video a recipient is able to view in response to correspondence between the temporal tags and the identification of said recipient, as well as a persistence value and overlap threshold;
- wherein said temporal tags are utilized in combination with an overlap threshold and a persistence value when determining which portions of said video a recipient is able to subsequently view;
- wherein said persistence value establishes length of video played for the recipient after persons identified by facial or voice recognition are out of a scene;
- and
- wherein said overlap threshold is configured to increase persistence by making video sections contiguous which have close temporal proximity and the same identified persons;
- wherein portions of said video are configured for the recipient to view in response to:
- (a) automatically distributing said video and said associated said temporal tags to a plurality of recipients wherein access to locations in said video by a said recipient is determined in response to correspondence between the temporal tags and the identification of said recipient;
- or (b) trimming said video into trimmed video clips based on correspondence between said temporal tags and the identification of a said recipient and automatically distributing the trimmed video clips to said recipient;
- wherein an appearance reset threshold is configured with a longer time duration than said persistence value, in which said appearance reset threshold establishes length of time an identified person must be absent from a video segment before considering this position in the video as the end of their presence.
14 more elements
A method of processing a video, comprising:
- inputting a video into a computer device;
- identifying one or more persons in said video in response to facial recognition;
- identifying one or more persons in said video in response to voice recognition of persons speaking within said video;
- generating temporal tags for locations in said video at which persons contained in a contact library are identified in response to using a combination of said facial recognition and said voice recognition;
- and
- storing said temporal tags within said video, or in a file associated with said video;
- wherein said temporal tags control which portions of said video a subsequent recipient of said video is able to view based on a correspondence between said temporal tags and the identification of said recipient;
- wherein said temporal tags are utilized in combination with an overlap threshold and a persistence value when determining which portions of said video a recipient is able to subsequently view;
- wherein said persistence value establishes length of video played for the recipient after persons identified by facial or voice recognition are out of a scene;
- and
- wherein said overlap threshold is configured to increase persistence by making video sections contiguous which have close temporal proximity and the same identified persons;
- further comprising wherein an appearance reset threshold which is configured with a longer time duration than said persistence value wherein said appearance reset threshold establishes length of time an identified person must be absent from a video segment before considering this position in the video as the end of their presence.
9 more elements
Dependencies are inferred from the claim text.