Table 5.
Model performance Feature Extractor loss optimal-dimension top layer. Bold indicates the best performing method, and grey highlights results that are not statistically significantly different from the best.
| MAP@R | ||||||
|---|---|---|---|---|---|---|
| Faces | Flanks | |||||
| Architecture | Lions | Chimps | Panda | Nyala | Zebra | Tiger |
| VGG-11 | 17.7 ± 1 | 13.8 ± 3 | 32.0 ± 2 | 11.2 ± 0 | 16.7 ± 2 | 22.8 ± 1 |
| ResNet-18 | 19.0 ± 1 | 11.2 ± 2 | 26.3 ± 1 | 9.9 ± 2 | 19.0 ± 0 | 24.6 ± 4 |
| DenseNet-201 | 20.8 ± 1 | 11.6 ± 3 | 31.1 ± 1 | 11.0 ± 2 | 16.0 ± 1 | 23.2 ± 3 |