Age Estimation on Adience
79.55Exact AccuracyMGA (ResNet-50)
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| MGA (ResNet-50)Param #=24M2018.09 | 79.55 | 89.74 | |
| ResNet-50Param #=23.6M2018.09 | 76.82 | 87.95 | |
| MGAParam #=2M2018.09 | 75.03 | 87.8 | |
| RoR-34Param #=21.8M2018.09 | 66.74 | 97.38 | |
| AttentionParam #=138M2018.09 | 61.8 | 95.1 | |
| Ordinal Hyperplane Network (Full model)2017.08 | 56.01 | 91.42 | |
| Ordinal Hyperplane Network (w/o LKL)loss=without LKL2017.08 | 54.52 | 91.29 | |
| Cumulative Attributefeatures=VGG face verification features, re-trained=true2017.08 | 52.34 | 89.34 | |
| Deep CNN2017.08 | 50.7 | 84.7 | |
| Ordinal Hyperplane Network (w/o Lhyper)loss=without Lhyper2017.08 | 49.46 | 82.99 | |
| LBP+FPLBP+Dropout 0.82017.08 | 45.1 | 79.5 |