An Exploration of Facial Expression Tracking in Affective HCI

The Significance of Textures for Affective Interfaces

“…With respect to the reactions of both types of face judges considered the movements of the virtual and real faces to be in agreement for 7 out of the 15 pairs of faces. In their conclusion Ward et al 2003, suggest that the failure of the facial tracking software is to blame for the results. However, in presenting their work, Ward et al suggested that participants in some cases seemed to be recognizing emotions in the virtual faces constructed by the facial tracking system that they had failed to notice on the films of real human faces [2].…”

“…Participants were 78.6% successful at recognising the correct emotion on the photographs but only 62.2% successful with the virtual heads. However, on closer inspection of the data it can be seen that fear and disgust had much lower scores for the virtual heads than for the other expressions and without these anomalies it may be possible that results would have shown no significant difference Ward et al [33] carried out an investigation monitoring the facial movements of participants engaged in a web-based task. Participants were asked to complete an online test that contains one surprise event and FaceStation 1.2, facial tracking software was used to track their faces as they completed the task.…”

“…The captured data from this was then transferred to create the movements of a virtual face. 'Judges' were asked to observe the footage of the real faces completing the tasks and of the virtual faces that had been created by the tracking software [33]. With respect to the reactions of both types of face judges considered the movements of the virtual and real faces to be in agreement for 7 out of the 15 pairs of faces.…”