Table 4 Number and percentage of selected features from the fused descriptor
From: Driver head pose estimation using efficient descriptor fusion
| Â | Â | 3 classes for pitch-SVM | 3 classes for yaw-SVM | ||||
|---|---|---|---|---|---|---|---|
Descriptor | TN | SN | FiD | DiF | SN | FiD | DiF |
SF | 450 | 263 | 58 % | 66 % | 275 | 61 % | 69 % |
HOG | 90 | 62 | 68 % | 16 % | 70 | 78 % | 18 % |
SURF | 256 | 0 | 0 % | 0 % | 11 | 4 % | 2 % |
Haar | 1024 | 75 | 7 % | 18 % | 44 | 4 % | 11 % |