Invention Grant
- Patent Title: Model-driven learning for video analytics
-
Application No.: US16006440Application Date: 2018-06-12
-
Publication No.: US10657377B2Publication Date: 2020-05-19
- Inventor: Raghuraman Gopalan
- Applicant: AT&T Intellectual Property I, L.P.
- Applicant Address: US GA Atlanta
- Assignee: AT&T Intellectual Property I, L.P.
- Current Assignee: AT&T Intellectual Property I, L.P.
- Current Assignee Address: US GA Atlanta
- Agency: Guntin & Gust, PLC
- Agent Robert Gingher
- Main IPC: G06K9/00
- IPC: G06K9/00 ; H04N21/234 ; G06K9/66 ; G06K9/62

Abstract:
Aspects of the subject disclosure may include, for example, receiving a plurality of media items, each including an object having a plurality of variables, performing adaptive analysis and transfer analysis on the media items, generating a model based on both the adaptive analysis and the transfer analysis and recognizing, using the model, the object in media content. The model may also provide a description of at least one of the variables of the object in the media content, and/or how that variable relates to the object. The variables may be known in the media items and unknown in the media content. The adaptive analysis and/or the transfer analysis may be performed iteratively and/or simultaneously. The variables may include an intensity of the object, a viewpoint from which the object is observed, an illumination of the object, a blur of the object, and/or other variables. Other embodiments are disclosed.
Information query