Table 13.
Comparison of classification accuracy on KDEF-O with existing models
| Reference | Accuracy (%) on non-occluded faces | Accuracy (%) on occluded faces |
|---|---|---|
| Dong et al. (2019) | 94.56 | 91.42 |
| Liu et al. (2021) | 91.30 | – |
| Proposed CES-CNN | 94.42 | 92.86 |
Bold represents the better accuracy value of the proposed model