Table 7 Comparison with state-of-the-art
From: Predicting the Sixteen Personality Factors (16PF) of an individual by analyzing facial features
Work | Year | Method used | Facial features used | Personality type assessed | Highest prediction accuracy |
|---|---|---|---|---|---|
Setyadi et al. [4] | 2015 | ANNs | Distance between two corners of the eyes, high of eyes, ratio of the mouth width and nose, the width ratio of two eyes, and thickness of lower lip | four temperaments | 65–68% (for sanguine and melancholic temperaments) |
Teijeiro-Mosquera et al. [20] | 2015 | CERT | CERT—four sets of behavioral cues | FFM | 60–65% (for extraversion) |
Gavrilescu [21] | 2015 | FFNN | FACS—27 AUs | FFM | 75% (for neuroticism, openness to experience, and extraversion) |
Chin et al. [22] | 2013 | EM | Marker-based tracking | MBTI | 60% |
Zhang et al. [23] | 2017 | CNNs | CNN features | 16PF | > 80% for rule consciousness and tension |
Current work | 2017 | FFNNs | FACS—27 AUs | 16PF | 67% (random datasets for training and testing) 70–72% (emotion induced datasets for training and random datasets for testing) > 80% for warmth, emotional stability, liveliness, social boldness, sensitivity, vigilance, and tension |