Research Article
Multilayer Convolutional Feature Aggregation Algorithm for Image Retrieval
Table 2
mAP values of different layer features in the fine-tune model on two datasets.
| ā | Oxford5k | Paris6k | Pool3 | Pool4 | Pool5 | Pool3 | Pool4 | Pool5 |
| CroW | 46.0 | 67.3 | 77.2 | 60.6 | 75.3 | 82.0 | ML-RCroW | 48.9 | 69.8 | 79.3 | 64.1 | 78.3 | 81.7 | CroW+QE | 50.3 | 71.0 | 80.0 | 67.1 | 79.8 | 85.4 | ML-RCroW+QE | 54.8 | 75.6 | 83.4 | 69.8 | 82.2 | 85.5 |
|
|