Table 2.
True Class | ||
Predicted Class | PS | WS |
PS | 549 | 78 |
WS | 115 | 268 |
Of 1009 test cases, 80.9% were correctly classified by the SVM-RBF classifier. The confusion matrix shows that accuracy of classification for poorly segmented (PS) objects was somewhat better than that for well-segmented (WS) objects.