Age group identification using gaze-guided feature extraction

When an observer identifies the age group of a subject in an image, the observer’s gaze focuses on a region containing an informative feature. Our aim is to improve the accuracy of age group identification by extracting features from the regions where an observer’s gaze converges. Existing studies have analysed observer gaze on facial images, but not on images containing the subject’s whole body. Here, we analysed which regions of the whole-body image an observer’s gaze focused on while the observer performed this task. The experimental results revealed that an observer’s gaze is drawn to the head of the subject regardless of the subject’s age group. They also revealed that the gaze-guided feature extraction in deep learning and machine learning improved the accuracy of age group identification.