Table 5.
The performance of different regression models for amount estimation (Unit: %).
| Regression Model | Container-Independent | Container-Dependent | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Can | Bottle | Handleless Mug | Handled Mug | |||||||
| MPE | MAPE | MPE | MAPE | MPE | MAPE | MPE | MAPE | MPE | MAPE | |
| Linear | 28.96 | 49.53 | 27.73 | 48.51 | 20.65 | 44.22 | 24.12 | 43.16 | 24.44 | 44.53 |
| Gaussian | 15.72 | 44.45 | 18.58 | 45.80 | 23.13 | 44.20 | 11.19 | 38.66 | 11.64 | 41.10 |
| SVM-linear 1 | 12.68 | 40.06 | 5.65 | 29.53 | 9.65 | 34.28 | 7.14 | 38.94 | 16.28 | 46.69 |
| SVM-Poly 2 | 24.51 | 47.75 | 24.97 | 47.49 | 22.24 | 45.45 | 19.26 | 41.14 | 19.91 | 42.74 |
| SVM-RBF 3 | 20.66 | 45.93 | 18.86 | 44.25 | 14.01 | 38.92 | 14.93 | 40.86 | 19.96 | 45.80 |
1 SVM-linear: SVM regression model with linear kernel function; 2 SVM-Poly: SVM regression model with polynomial kernel function; 3 SVM-RBF: SVM regression model with RBF kernel function.