A performance comparison of eight commercially available automatic classifiers for facial affect recognition
[FULL TEXT READ at the publisher] Dupre, D., Krumhuber, E. G., Kuster, D., & McKeown, G. J. (2020). 'A performance comparison of eight commercially available automatic classifiers for facial affect recognition.' PLOS ONE, 15(4), e0231968. doi:10.1371/journal.pone.0231968. THE ONLY PEER-REVIEWED HEAD TO HEAD OF COMMERCIAL EMOTION CLASSIFIERS AND THE CONSUMER SPINE OF THE ARTICLE. THE EIGHT SYSTEMS, with the paper's own names and versions: Affectiva's Affdex (SDK v3.4.1, HOG features plus SVM, distributed by Affectiva and by iMotions); CrowdEmotion's FaceVideo (API v1.0, CNN); Emotient's Facet (SDK v6.3, distributed by iMotions, and the paper states Emotient was acquired by Apple in 2017 and that Facet 'is no longer commercially available, but existing licences are still supported by iMotions'); Microsoft's Cognitive Services Face (API v1.0, Azure, first released 2015); MorphCast's EmotionalTracking (SDK v1.0); Neurodata Lab's EmotionRecognition (API v1.0); VicarVision's FaceReader (software v7.0, 'now distributed by Noldus', Active Appearance Models plus CNN, default settings); and Visage Technologies' FaceAnalysis (SDK v1.0). All eight output the six basic emotions, rescaled to odds ratios for comparability. STIMULI, 937 dynamic videos from two databases: BU-4DFE, POSED, 467 videos from 78 individuals instructed by a psychologist to portray each of the six emotions, with anger 78, disgust 78, fear 78, happiness 78, sadness 78 and surprise 77, mean duration 4.05 seconds, undergraduates, graduates and faculty at a New York state university aged 18 to 45; and UT-Dallas, SPONTANEOUS, 470 videos elicited by a ten-minute film compilation, students aged 18 to 25, very unevenly distributed with anger 3, disgust 119, fear 13, happiness 196, sadness 38 and surprise 101, mean duration 6.11 seconds. The authors describe the subjects of both in the same words: 'The majority of
[FULL TEXT READ at the publisher] Dupre, D., Krumhuber, E. G., Kuster, D., & McKeown, G. J. (2020). 'A performance comparison of eight commercially available automatic classifiers for facial affect recognition.' PLOS ONE, 15(4), e0231968. doi:10.1371/journal.pone.0231968. THE ONLY PEER-REVIEWED HEAD TO HEAD OF COMMERCIAL EMOTION CLASSIFIERS AND THE CONSUMER SPINE OF THE ARTICLE. THE EIGHT SYSTEMS, with the paper's own names and versions: Affectiva's Affdex (SDK v3.4.1, HOG features plus SVM, distributed by Affectiva and by iMotions); CrowdEmotion's FaceVideo (API v1.0, CNN); Emotient's Facet (SDK v6.3, distributed by iMotions, and the paper states Emotient was acquired by Apple in 2017 and that Facet 'is no longer commercially available, but existing licences are still supported by iMotions'); Microsoft's Cognitive Services Face (API v1.0, Azure, first released 2015); MorphCast's EmotionalTracking (SDK v1.0); Neurodata Lab's EmotionRecognition (API v1.0); VicarVision's FaceReader (software v7.0, 'now distributed by Noldus', Active Appearance Models plus CNN, default settings); and Visage Technologies' FaceAnalysis (SDK v1.0). All eight output the six basic emotions, rescaled to odds ratios for comparability. STIMULI, 937 dynamic videos from two databases: BU-4DFE, POSED, 467 videos from 78 individuals instructed by a psychologist to portray each of the six emotions, with anger 78, disgust 78, fear 78, happiness 78, sadness 78 and surprise 77, mean duration 4.05 seconds, undergraduates, graduates and faculty at a New York state university aged 18 to 45; and UT-Dallas, SPONTANEOUS, 470 videos elicited by a ten-minute film compilation, students aged 18 to 25, very unevenly distributed with anger 3, disgust 119, fear 13, happiness 196, sadness 38 and surprise 101, mean duration 6.11 seconds. The authors describe the subjects of both in the same words: 'The majority of
Written from the fULL TEXT READ at the publisher.
What our sources record
Recorded when this source was checked, figures included, with page references and quotation marks removed.
Dupre, D., Krumhuber, E. G., Kuster, D., & McKeown, G. J. (2020). A performance comparison of eight commercially available automatic classifiers for facial affect recognition. PLOS ONE, 15(4), e0231968. doi:10.1371/journal.pone.0231968