TABLE 5.
Results of the proposed MagNet and existing algorithms on the proposed ID Age nder databases using frames/images as input.
| Database | Features | EER (%) | ACER (%) |
|---|---|---|---|
| Snapchat | LBP Määttä et al. (2011) | 27.1 ± 4.3 | 27.3 ± 4.1 |
| LPQ Ojansivu and Heikkilä (2008) | 28.7 ± 4.0 | 30.4 ± 3.8 | |
| BSIF Kannala and Rahtu (2012) | 30.2 ± 7.0 | 30.2 ± 6.9 | |
| VGG-16 Simonyan and Zisserman (2015) | 17.7 ± 2.4 | 18.4 ± 2.3 | |
| GoogLeNet Szegedy et al. (2015) | 28.1 ± 5.1 | 29.1 ± 4.9 | |
| S-MIL Li et al. (2020a) | 16.9 ± 3.6 | 18.2 ± 2.7 | |
| XceptionNet Rossler et al. (2019) | 19.7 ± 4.7 | 23.6 ± 3.1 | |
| ResNet-18 Kumar et al. (2020) | 30.0 ± 5.9 | 31.6 ± 5.7 | |
| Proposed (MagNet) | 18.0 ± 0.4 | 17.6 ± 0.3 | |
| Identity Morphing | LBP Määttä et al. (2011) | 0.6 ± 0.2 | 0.9 ± 0.1 |
| LPQ Ojansivu and Heikkilä (2008) | 6.1 ± 0.3 | 6.2 ± 0.2 | |
| BSIF Kannala and Rahtu (2012) | 6.2 ± 0.4 | 6.2 ± 0.2 | |
| VGG-16 Simonyan and Zisserman, (2015) | 4.7 ± 1.1 | 9.7 ± 1.0 | |
| GoogLeNet Szegedy et al. (2015) | 12.3 ± 2.1 | 11.5 ± 0.9 | |
| S-MIL Li et al. (2020a) | 9.4 ± 1.2 | 11.7 ± 1.8 | |
| XceptionNet Rossler et al. (2019) | 7.9 ± 2.4 | 9.1 ± 1.1 | |
| ResNet-18 Kumar et al. (2020) | 8.5 ± 1.8 | 10.6 ± 1.2 | |
| Proposed (MagNet) | 0.0 ± 0.0 | 0.2 ± 0.0 | |
| FaceApp | LBP Määttä et al. (2011) | 1.3 ± 0.8 | 2.7 ± 0.7 |
| LPQ Ojansivu and Heikkilä (2008) | 1.2 ± 0.4 | 1.3 ± 0.3 | |
| BSIF Kannala and Rahtu (2012) | 30.3 ± 4.4 | 30.5 ± 4.5 | |
| VGG-16 Simonyan and Zisserman (2015) | 18.3 ± 2.5 | 21.4 ± 2.3 | |
| GoogLeNet Szegedy et al. (2015) | 23.7 ± 2.1 | 24.5 ± 2.7 | |
| S-MIL Li et al. (2020a) | 8.6 ± 1.8 | 12.3 ± 1.2 | |
| XceptionNet Rossler et al. (2019) | 10.2 ± 3.6 | 14.7 ± 1.8 | |
| ResNet-18 He et al. (2016) | 16.2 ± 3.3 | 14.8 ± 1.6 | |
| Proposed (MagNet) | 0.4 ± 0.7 | 2.5 ± 0.4 |
Result of the best performing algorithm is highlighted in bold.