Automated detection of smiles as discrete episodes

Abstract

Smiling is a spontaneous facial expression occurring throughout everyday life, which varies largely between individuals. 1 While the interpretation of smiling may appear straightforward, it is actually one of the most complex facial expressions, and can be ambiguous. 2 Not only can smiles have different forms and meanings, but they are also found in different situations and as a consequence of different eliciting factors. 3,4 Smile analysis in dentistry has largely focused on static images. 5 Nonetheless, more recently, there has been a paradigm shift in

What’s in a Smile? The features of Indigenous Fijians’ and NZ Europeans’ smiles

“…The onset of a smiling episode was de ned as activation of both AU6 and AU12 at or above 0.5 and 1.5, respectively. These thresholds were optimised in our previous validation study (Mohammed et al, 2022). The offset of the smiling episode was de ned as subthreshold activation of either action unit for longer than two seconds (60 frames).…”

“…A 0/1 score was assigned to AU25, with 1 indicating that the participant's teeth were visible. For each smiling episode, we calculated the frequency, time of onset, duration, and mean activation of AU6 and AU12 as well as the proportion of time the participant's teeth were visible (Mohammed et al, 2022).…”

“…Another limitation of this study was that the automated smiling analysis software used in the current study was previously trained and calibrated on a Caucasian sample (Mohammed et al, 2022). While there is limited evidence for ethnic differences in facial musculature, there are anatomical variations between individuals (Goodmurphy & Ovalle, 1999) as a function of gender and age (Silva et al, 2010).…”

“…A compilation of three amusing videos was used as a smile trigger. The videos included an episode of Mr Bean, which has been used in previous research on smiles (Juckel et al, 2011;Zweyer et al, 2004) and other two videos (Mohammed et al, 2022). The entire video clip was 4 minutes and 33 seconds in length.…”

Smile dynamics: A comparative study of Fijians and NZ Europeans

“…The videos included an episode of Mr Bean, which has been used in previous research on smiles 17,18 and other two videos. 19 The entire video clip was 4 min and 33 s in length. Facial lighting was optimised for each participant by using a ring light mounted on the back of the display monitor.…”