Patent · US Active

Techniques for identifying objects from video content

US11468675B1 · kind B1 · utility

1Cited by
2References
20Claims
0Family size

Assignee

Inventor

Key dates

Filing dateDec 3, 2019
Grant dateOct 11, 2022
Priority date
Expiry dateDec 27, 2040

Classification

  • Technology area (CPC G)Physics
  • CPC primaryG06V10/82
  • WIPO fieldComputer technology
  • WIPO sectorElectrical engineering

Abstract

Systems, methods, and computer storage media are described herein for identifying objects within video content. Utilizing an image recognition technique (e.g., a previously trained neural network) an object may be detected within a subset of the video frames of a scene of the video content. Metadata may be generated comprising a set of object attributes (e.g., an object image, a URL from which the item may be procured, etc.) associated with the object detected within the subset of video frames. A request for object identification may subsequently be received while a user is watching the video content. A run time of the request may be utilized to retrieve and display one or more object attributes (e.g., an object image with an embedded hyperlink) corresponding to an object appearing in the scene.

Source: USPTO / EPO open patent data. Objective bibliographic and citation counts.