Table 4.
The best performance of gesture spotting using different machine learning models.
| Machine Learning Model | Window Size (Samples) | Overlap (%) | Metric | Gesture | Overall | ||||
|---|---|---|---|---|---|---|---|---|---|
| Fetch | Lift | Sip | Drop | Release | |||||
| ADA | 16 | 50 | Sensitivity (%) | 83.26 | 89.74 | 93.10 | 90.62 | 76.58 | 86.66 |
| Precision (%) | 85.64 | 91.55 | 93.83 | 89.88 | 85.26 | 89.23 | |||
| DT | 16 | 25 | Sensitivity (%) | 89.72 | 84.08 | 92.24 | 86.98 | 80.79 | 86.76 |
| Precision (%) | 84.03 | 91.92 | 94.66 | 91.61 | 87.53 | 89.95 | |||
| RF | 16 | 50 | Sensitivity (%) | 89.35 | 88.70 | 94.75 | 90.61 | 87.44 | 90.17 |
| Precision (%) | 91.63 | 95.27 | 95.35 | 93.98 | 87.80 | 92.80 | |||
| NB | 16 | 25 | Sensitivity (%) | 57.35 | 77.01 | 90.46 | 88.46 | 54.78 | 73.61 |
| Precision (%) | 71.02 | 90.93 | 88.23 | 64.08 | 71.51 | 77.15 | |||
| k-NN | 16 | 50 | Sensitivity (%) | 78.10 | 83.04 | 95.69 | 85.60 | 71.58 | 82.80 |
| Precision (%) | 84.53 | 86.19 | 85.28 | 82.05 | 86.04 | 84.82 | |||
| SVM | 16 | 25 | Sensitivity (%) | 76.87 | 87.14 | 96.28 | 90.62 | 75.80 | 85.34 |
| Precision (%) | 83.30 | 95.64 | 94.58 | 92.70 | 78.41 | 88.93 | |||
ADA: AdaBoost; DT: decision tree; RF: random forest; NB: Naïve Bayes; k-NN: k-nearest neighbors; SVM: support vector machine.