Research Article
Script Identification from Printed Indian Document Images and Performance Evaluation Using Different Classifiers
Table 4
Statistical performance analysis and weighted average of the measuring parameters are shown here.
| Classifier | KS | MAE | RAE | TP rate | FP rate | Precision | Recall | -measure |
| BayesNet | 0.9631 | 0.007 | 3.9702 | 0.967 | 0.004 | 0.969 | 0.967 | 0.967 | LibLINEAR | 0.9729 | 0.0048 | 2.7033 | 0.976 | 0.003 | 0.977 | 0.976 | 0.976 | MLP | 0.9828 | 0.007 | 3.9239 | 0.985 | 0.002 | 0.985 | 0.985 | 0.985 | RBFNetwork | 0.9509 | 0.0101 | 5.6777 | 0.956 | 0.005 | 0.960 | 0.956 | 0.957 | Simple Logistic | 0.9877 | 0.0059 | 3.3448 | 0.989 | 0.001 | 0.989 | 0.989 | 0.989 | FURIA | 0.9277 | 0.0198 | 7.2849 | 0.941 | 0.013 | 0.941 | 0.941 | 0.940 | PART | 0.9065 | 0.0177 | 9.9616 | 0.917 | 0.010 | 0.917 | 0.917 | 0.916 | NBTree | 0.9312 | 0.0159 | 8.978 | 0.939 | 0.008 | 0.940 | 0.939 | 0.939 | Random Forest | 0.9729 | 0.0248 | 13.9833 | 0.976 | 0.003 | 0.976 | 0.976 | 0.975 |
|
|
Parameters considered: KS: Kappa statistics, MAE: mean absolute error, RAE: relative absolute error and TP rate, FP rate, precision, recall, -measure.
|