Table 10 Performance Metrics of SignaryNet over Benchmark datasets
Metrics | SIW-13 (Printed) (321 Line images) | CVSI-2015 (Scene text) (50 Word images) | Wikipedia Ancient Tamil script (41 raw images) |
|---|---|---|---|
Training | 0.9916 | 0.9906 | 0.9600 |
Validation | 0.9880 | 0.9969 | 0.9960 |
Testing | 0.9816 | 0.9780 | 0.9890 |
Precision | 0.9882 | 0.9872 | 0.9935 |
Recall | 0.9916 | 0.9901 | 0.9953 |
F-score | 0.9900 | 0.9886 | 0.9944 |
SRR | 0.9816 | 0.9780 | 0.9890 |
NPV | 0.9070 | 0.8805 | 0.9200 |
Specificity (TNR) | 0.8736 | 0.8500 | 0.9047 |
Sensitivity (TPR) | 0.9916 | 0.9901 | 0.9953 |
FNR | 0.0084 | 0.0099 | 0.0053 |
FDR | 0.0118 | 0.0128 | 0.0065 |
FPR | 0.1264 | 0.1500 | 0.0958 |
AUC | 0.9326 | 0.9201 | 0.9500 |
