An object detection device is provided with a detector for detecting an object to be detected depicted in an assessment image on the basis of a feature value of the object which has already been learned using a training image.
Disclosed is an object detection device for detecting an object to be detected depicted in an assessment image on the basis of a feature value of the object which has already been learned using a training image, wherein the object detection device is provided with a detector (2) for running strong classifiers (21m) in order of low determination precision. If the strong classifiers (21m) determine that the object to be detected is depicted in the assessment image, processing is continued, and if the strong classifiers (21m) determine that the object is not depicted in the assessment image, the object is treated as not detected without running strong classifiers (21m) of higher determination precision than the first-mentioned strong classifiers (21m); wherein into each of the strong classifiers (21m) is input a determination result of a strong classifier (21m) of lower determination precision than the first-mentioned strong classifier (21m), and on the basis of a plurality of estimates and input determination results, it is determined whether or not the object is depicted in the assessment image.