Table 1 Comparison of performance of Fisher Vectors, VLAD, and VLAD-FV
From: Refining deep convolutional features for improving fine-grained image recognition
Encoding methods | Cub (%) | Air (%) | Cars (%) | Dogs (%) |
|---|---|---|---|---|
Fisher Vectors | 79.0 | 82.3 | 83.7 | 64.1 |
VLAD | 80.8 | 83.0 | 87.1 | 65.8 |
VLAD-FV | 82.0 | 84.1 | 88.5 | 67.5 |