Adeia Guides Inc.

Rovi Guides, Inc.

Anthony Friede

Johan Kölhi

A first content type of a first media stream being output is determined, and at least one first classification model is selected based on the first content type. Objects are identified in the first media stream using the at least one first classification model and information related to the objects is generated for presentation. A change in output of media streams from the first media stream to a second media stream may then be detected. A second content type of the second media stream is then determined and at least one second classification model is selected based on the second content type. Objects are then identified in the second media stream using the at least one second classification model and information related to the objects in generated for presentation.

Systems and methods for leveraging machine learning to enable user-specific real-time information services for identifiable objects within a video stream

Systems and methods are described for using a biometric response profile to predict a biometric response of a user to a future event. An event occurring in video content is determined. Biometric data of a user viewing the video content is received, the biometric data comprising a biometric response of the user to the event. A biometric response profile of the user indicating a predicted biometric response of the user to a future event corresponding to the event is generated based on the biometric data and the event.

Methods and systems for generating a biometric response profile

At least one high-quality image of a speaker is captured. A low network quality condition may be detected between a client device and a video service node. In response to detecting the low network quality condition, a data stream comprising changes to the high-quality image of the speaker needed to recreate a representation of the speaker is generated. Transmission of the video stream of the speaker between the client device of the speaker and the video service node is stopped and, simultaneously, transmission of the data stream is begun. A digital twin of the speaker is then generated for display at the client device based on the data stream and the high-quality image of the speaker.

Growing community of inventors

Upplands Väsby, Sweden

Anthony Friede