Table 7 SVM models performances measures using fivefold CV (with and without data augmentation)
From: Classification of facial paralysis based on machine learning techniques
Classifier | #1 Smiling | #2 Closing eyes | #3 Raising eyebrows | #4 Blowing cheeks | #5 Whistling | |||||
|---|---|---|---|---|---|---|---|---|---|---|
Augmentation? | No | Yes | No | Yes | No | Yes | No | Yes | No | Yes |
Accuracy % | 90 ± 7 | 98 ± 1 | 86 ± 4 | 98 ± 1 | 73 ± 8 | 94 ± 3 | 82 ± 8 | 93 ± 3 | 87 ± 5 | 96 ± 4 |
Precision % | 90 ± 10 | 98 ± 2 | 81 ± 9 | 98 ± 1 | 71 ± 15 | 94 ± 4 | 87 ± 5 | 93 ± 3 | 90 ± 6 | 95 ± 4 |
Sensitivity % | 89 ± 11 | 98 ± 2 | 78 ± 10 | 98 ± 1 | 70 ± 10 | 93 ± 4 | 84 ± 5 | 93 ± 4 | 87 ± 4 | 96 ± 3 |
F1-score % | 88 ± 12 | 98 ± 2 | 77 ± 9 | 98 ± 1 | 68 ± 14 | 93 ± 4 | 84 ± 5 | 93 ± 4 | 85 ± 6 | 95 ± 4 |