Electronic device, information processing apparatus, method for inference, and program for inference

An information processing apparatus comprising:

• an acquiring unit configured to acquire at least one image corresponding to a view and a gaze line of a subject, the gaze line being directed to the view;
• a controller configured to estimate an alertness level of the subject based on the at least one image and the gaze line;
• and
• an output unit configured to output the alertness level,
• wherein the controller is configured to serve as a first inference unit capable of creating a gaze-line prediction map that is based on the at least one image and that represents a frequency of finding the gaze line in the at least one image, the first inference unit being constructed based on learning data obtained by machine learning regarding a relationship between a training image and a gaze line of a subject for training, the gaze line being directed to the training image, and as a second inference unit capable of inferring biological information regarding the alertness level of the subject based on the gaze-line prediction map and the gaze line of the subject, the second inference unit being constructed based on learning data obtained by machine learning regarding a relationship among a gaze-line prediction map for training inferred by the first inference unit based on the training image, the gaze line of the subject for training, the gaze line being directed to the training image, and biological information regarding an alertness level of the subject for training.

A method for inference, comprising the steps of:

• creating at least one image corresponding to a view by image capturing;
• detecting a gaze line of a subject, the gaze line being directed to the view;
• and
• estimating an alertness level of the subject based on the at least one image and the gaze line,
• wherein the step of estimating includes
• a first inference step in which a gaze-line prediction map can be inferred based on the at least one image, the gaze-line prediction map representing a probability of finding the gaze line at each position in the at least one image, the first inference step being constructed based on learning data obtained by machine learning regarding a relationship between a training image and a gaze line of a subject for training, the gaze line being directed to the training image, and
• a second inference step in which biological information regarding the alertness level of the subject can be inferred based on the gaze-line prediction map and the gaze line of the subject, the second inference step being constructed based on learning data obtained by machine learning regarding a relationship among a gaze-line prediction map for training inferred in the first inference step based on the training image, the gaze line of the subject for training, the gaze line being directed to the training image, and biological information regarding an alertness level of the subject for training.

A non-transitory computer-readable recording medium including program for inference causing a computer to serve as

• an imaging unit configured to create at least one image corresponding to a view by image capturing;
• a gaze-line detector configured to detect a gaze line of a subject, the gaze line being directed to the view;
• and
• a controller configured to estimate an alertness level of the subject based on the at least one image and the gaze line,
• wherein the controller is configured to serve as
• a first inference unit capable of inferring a gaze-line prediction map that is based on the at least one image and that represents a probability of finding the gaze line at each position in the at least one image, the first inference unit being constructed based on learning data obtained by machine learning regarding a relationship between a training image and a gaze line of a subject for training, the gaze line being directed to the training image, and
• a second inference unit capable of inferring biological information regarding the alertness level of the subject based on the gaze-line prediction map and the gaze line of the subject, the second inference unit being constructed based on learning data obtained by machine learning regarding a relationship among a gaze-line prediction map for training inferred by the first inference unit based on the training image, the gaze line of the subject for training, the gaze line being directed to the training image, and biological information regarding an alertness level of the subject for training.