User recognition and gaze tracking in a video system

A system comprising:

• one or more processors;
• a display;
• and
• an image sensor configured to generate observational data of a first detected user within a field-of-view, the display and image sensor communicatively coupled to the one or more processors,
• wherein the one or more processors are configured to:
• periodically track a location of the first detected user based on the observational data;
• store the periodically tracked locations of the first detected user as historical data;
• determine user traits for the first user including where the first user is likely to be or where the user is likely to return to when the user is out of the field-of-view based on the historical data;
• determine when the first detected user leaves the field-of-view based on the observational data;
• detect when a second user enters the field-of-view;
• and
• determine a confidence score indicating a probability that the second detected user is the first detected user based on the determined user traits for the first user.

A computer-implemented method comprising:

• receiving, by one or more processors, observational data of a first detected user from one or more image sensors;
• periodically tracking a location of the first detected user based on the observational data;
• storing the periodically tracked locations of the first detected user as historical data;
• determining user traits for the first user including where the first user is likely to be or where the user is likely to return to when the user is out of the field-of-view based on the historical data;
• determining when the first detected user leaves the field-of-view based on the observational data;
• detecting when a second user enters the field-of-view;
• and
• determining a confidence score indicating a probability that the second detected user is the first detected user based on the determined user traits for the first user.

A non-transitory computer-program product tangibly embodied in a machine-readable non-transitory storage medium that includes instructions configured to cause one or more processors to:

• receive observational data of a first detected user from one or more image sensors;
• periodically track a location of the first detected user based on the observational data;
• store the periodically tracked locations of the first detected user as historical data;
• determine user traits for the first user including where the first user is likely to be or where the user is likely to return to when the user is out of the field-of-view based on the historical data;
• determine when the first detected user leaves the field-of-view based on the observational data;
• detect when a second user enters the field-of-view;
• and
• determine a confidence score indicating a probability that the second detected user is the first detected user based on the determined user traits for the first user.