From: Development and application of emotion recognition technology — a systematic literature review
No | Author & year | Feature extraction | N | Emotions | Test sample | Pattern recognition methods | Accuracy |
---|---|---|---|---|---|---|---|
1 | Madrigal Garcia (2018) | Facial image | 34 | Sadness, Fear | Patients at risk of clinical deterioration | Pearson, Chi-square test, T-test, Cluster analysis, 1LR | NA |
2 | Xin Chen (2019) | Facial image | 49 | NA | Inpatients | CNN | Training is 92%, Test is 83% |
3 | Isabelle Chiu (2015) | Facial stimulation | 49 | Anger, Disgust, Fear, Happiness, Sadness, Astonished | Healthy individuals aged between 52 and 79 | 2GLMM | NA |
4 | Kowallik Andrea E (2021) | Facial action unit | 55 | Anger, Disgust, Fear, Happiness, Sadness, Astonished | Patients with different degrees of autism | Multinomial Logistic Regression Algorithm | F(1, 53)= 2.428, p = 0.125, ηp2 = 0.044 |
5 | Onyema, Edeh Michael (2021) | Facial image | NA | Anger, Disgust, Fear, Happiness, Sadness, Astonished, Neutral | FER 2013 dataset | CNN | 70% |
6 | Masulli Paolo (2022) | Eye tracking data | 111 | Autism, Depression, Attention deficit | Psychiatric outpatients | PCA, Linear regression | NA |
7 | Parra-Dominguez (2022) | Eyebrows, eyes, mouth (angle, slope, Euclidean distance, perimeter of a closed shape) | 120 | NA | MEEI database | 5-fold cross validation, 3ANN | 90.25% |
8 | Toshiya Akiyama (2022) | Facial action unit and image | 71 | Anger, Disgust, Fear, Happiness, Sadness, Astonished, Neutral | Schizophrenics | Multi-task CNN | 66.29% |
9 | Jiayu Ye (2022) | Facial action unit and image | 164 | Anger, Disgust, Fear, Happiness, Sadness, Astonished, Neutral | Depression patients | Spearman correlation analysis, Random forest, 4LR-RFE | 41.7% |
10 | Muhammad Munsif (2022) | Facial image | 70 | Normal, Happy, Angry, sad | KEDF dataset | CNN | Training is 96%, Test is 97% |
11 | Kuttenreich, Anna-Maria (2022) | Facial image | 60 | Anger, Disgust, Fear, Happiness, Sadness, Astonished | Patients with facial linkage after paralysis | NA | FER accuracy 67.7 ± 11.3%, AER accuracy 67.7 ± 11.3% |
12 | Yiming Fan (2022) | Facial image | 84 | Anger, Pain, Tension, Happiness, Sadness, Astonished, Fatigue, Neutral | RAF-DB dataset, FER + dataset, Private dataset of stroke patients | CNN | FER+:88.21%, RAF-DB:89.44%, private datasets 99.81% |