Validating the AU-mapping recognition of emotion recognition software (OpenFace)

Since the beginning of Ekman's (1970) research, a lot of research has been done on emotions. According to Ekman there are 6 different emotional facial expressions: happiness, sadness, anger, fear,surprise, and disgust. Originally these emotions were coded by trained professionals. The current research therefore focuses on the question of whether emotional facial expressions can also be recognized correctly by facial recognition software (such as OpenFace). This is done by showing participants a selection of stimuli (3 second segments) after which they must score this segment on the 6 basic emotions of Ekman (1970). The current results indicate that facial recognition software derives the same kind of emotion as the participants do. This is an indication that facial recognition software could replace trained professionals. However, much research isstill needed about the quality of the software recognition compared to the recognition quality of a trained professional. In addition,...Show moreSince the beginning of Ekman's (1970) research, a lot of research has been done on emotions. According to Ekman there are 6 different emotional facial expressions: happiness, sadness, anger, fear,surprise, and disgust. Originally these emotions were coded by trained professionals. The current research therefore focuses on the question of whether emotional facial expressions can also be recognized correctly by facial recognition software (such as OpenFace). This is done by showing participants a selection of stimuli (3 second segments) after which they must score this segment on the 6 basic emotions of Ekman (1970). The current results indicate that facial recognition software derives the same kind of emotion as the participants do. This is an indication that facial recognition software could replace trained professionals. However, much research isstill needed about the quality of the software recognition compared to the recognition quality of a trained professional. In addition, research has also been conducted about the influence of moods on recognizing emotions. According to Bower's (1987) mood-congruence theory, people process a stimulus faster 4 and more attentively when it portrays the same mood. In addition, Steephen et al. (2020) argue that the intensity of recognition increases, when the stimulus contains the same emotion as the mood of the individual. This is exactly what has been studied. The results show that this is not the case, except for anxiety. This contradiction may be related to the study design, but also to the number of participants who participated in the study.Show less

Leiden University Student Repository

Documents

In Collections

Validating the AU-mapping recognition of emotion recognition software (OpenFace)