Table 6.
Classification F1-score comparison with state-of-the-art models.
| Model Names | Epoch | Batting | Boundary | Bowling | Close-Ups | Crowd | Mean |
|---|---|---|---|---|---|---|---|
| Minhas et al. | 1000 | 88% | 96% | 90% | 96% | 88% | 92% |
| Inception v3 | 50 | 96% | 99% | 98% | 100% | 97% | 98% |
| ResNet50 | 50 | 94% | 100% | 97% | 100% | 99% | 98% |
| VGGNET16 | 50 | 90% | 98% | 95% | 98% | 96% | 95% |
| VGGNET19 | 50 | 91% | 98% | 96% | 98% | 95% | 96% |
| SVM (64x64) | NA | 84% | 90% | 95% | 92% | 88% | 90% |
| Proposed Method | 50 | 96% | 100% | 98% | 100% | 100% | 99% |
Class wise F1-score in comparison with state-of-the-art models shows a robust performance.