Table 4.
Average precision of models using SVM and dimension of features
| Feature extractor | Average precision | Average precision | Feature dimension |
|---|---|---|---|
| (Linear SVM) | (SVM with RBF kernel) | ||
| PHOW | 80.24 | 85.16 | 4000 |
| LBP | 82.39 | 82.90 | 928 |
| PHOG | 49.04 | 62.60 | 680 |
| AlexNet Cropping | 88.28 | 88.12 | 4096/4096/3 |
| VGG-16 Cropping | 90.74 | 90.49 | 4096/4096/3 |
| GoogLeNet Cropping | 87.20 | 87.33 | 1000/1000 |
| NBI-Net Aug | 92.74 | 92.70 | 1024/128/3 |
The top results have been styled with bold and italic
The best results of comparative group are styled with italic