Research Article
A Novel Multicategory Defect Detection Method Based on the Convolutional Neural Network Method for TFT-LCD Panels
Table 6
Result of sensitivity and specificity results—modified dataset.
| Label in the model | Category | Specificity (%) | Sensitivity (%) | TPA (%) |
| 0 | Blue full-dark defect | 99.96 | 100 | 98.90 | 1 | Blue half-dark defect | 99.91 | 99.38 | 2 | Blue half-bright defect | 99.78 | 97.44 | 3 | Blue full-bright defect | 100 | 97.65 | 4 | Blue defect-free | 99.78 | 97.69 | 5 | Green full-dark defect | 100 | 99.39 | 6 | Green half-dark defect | 100 | 96.97 | 7 | Green half-bright defect | 100 | 100 | 8 | Green full-bright defect | 100 | 100 | 9 | Green defect-free | 99.73 | 100 | 10 | Red full-dark defect | 99.91 | 98.66 | 11 | Red half-dark defect | 100 | 98.06 | 12 | Red half-bright defect | 99.82 | 100 | 13 | Red full-bright defect | 100 | 99.41 | 14 | Red defect-free | 100 | 100 |
|
|