METHOD AND APPARATUS FOR ESTIMATING HUMAN EMOTION BASED ON ADAPTIVE IMAGE RECOGNITION USING INCREMENTAL DEEP LEARNING

摘要

According to one embodiment of the present invention, an apparatus for estimating an emotional state based on adaptive image recognition using incremental deep learning comprises: a face detection unit for detecting a facial region of a person from an image including the person who is a target for estimation of an emotional state; a body detection unit for detecting a body region of the person from the image including the person who is a target for estimation of an emotional state; a facial expression recognition unit for recognizing a facial expression by using an image of the detected facial region as an input; a behavior recognition unit for recognizing a behavior of the person by using an image of the detected body region as an input; and an emotional state determining unit for determining an emotional state of the person based on the recognized facial expression and behavior. Accordingly, it is possible to recognize an image resistantly to environmental changes since a recognition rate declining as the environment changes is quickly recovered with less effort and to gradually learn human behaviors and facial expressions with little data through incremental deep learning, and thus, it is possible to more precisely estimate an emotional state of a human being through a combination of two results.