Sight line position processing apparatus, image capturing apparatus, training apparatus, sight line position processing method, training method, and storage medium

A sight line position processing apparatus comprising at least one processor and/or at least one circuit which functions as:

• a detection unit configured to repeatedly detect a sight line position of a user;
• and
• a generation unit configured to generate a statistical sight line position by calculating a statistical value for a plurality of sight line positions detected by the detection unit, the generation unit calculating the statistical value using a higher number of sight line positions when a sight line position at a first distance from a center of a display is included than when a sight line position at a second distance shorter than the first distance is included.

A training apparatus comprising at least one processor and/or at least one circuit which functions as:

• a training unit configured to generate trained dictionary data by performing training based on a first plurality of eye area images each associated with information indicating a sight line position,
• wherein the training unit generates trained dictionary data that can be used for both left and right on the basis of horizontal symmetry of a right eye and a left eye in the first plurality of eye area images.

A sight line position processing method executed by a sight line position processing apparatus, comprising:

• repeatedly detecting a sight line position of a user;
• and
• generating a statistical sight line position by calculating a statistical value for a plurality of sight line positions detected by the detecting, wherein the statistical value is calculated using a higher number of sight line positions when a sight line position at a first distance from a center of a display is included than when a sight line position at a second distance shorter than the first distance is included.

A training method executed by a training apparatus, comprising:

• generating trained dictionary data by performing training based on a first plurality of eye area images each associated with information indicating a sight line position,
• wherein the generating generates trained dictionary data that can be used for both left and right on the basis of horizontal symmetry of a right eye and a left eye in the first plurality of eye area images.

A non-transitory computer-readable storage medium which stores a program for causing a computer to execute a sight line position processing method comprising:

• repeatedly detecting a sight line position of a user;
• and
• generating a statistical sight line position by calculating a statistical value for a plurality of sight line positions detected by the detecting, wherein the statistical value is calculated using a higher number of sight line positions when a sight line position at a first distance from a center of a display is included than when a sight line position at a second distance shorter than the first distance is included.

A non-transitory computer-readable storage medium which stores a program for causing a computer to execute a training method comprising:

• generating trained dictionary data by performing training based on a first plurality of eye area images each associated with information indicating a sight line position,
• wherein the generating generates trained dictionary data that can be used for both left and right on the basis of horizontal symmetry of a right eye and a left eye in the first plurality of eye area images.