Table 4 The ablation studies on VeRi-776 dataset (in %).
From: A vehicle re-identification framework based on the improved multi-branch feature fusion network
Method | VeRi-776 | |
|---|---|---|
mAP | Rank-1 | |
Baseline | 69.18 | 93.21 |
(a) Baseline + Local Branch1(no red dotted area) + CAM | 72.13 | 94.04 |
(b) Baseline + Local Branch2(no red dotted area) + CAM | 73.05 | 94.16 |
(c) Local Branch1(no red dotted area) + Local Branch2(no red dotted area and green dotted area) | 63.90 | 90.82 |
(d) Local Branch1(no red dotted area) + Local Branch2(no red dotted area) + Green dotted area | 69.57 | 91.66 |
(e) Local Branch1(no red dotted area) + Local Branch2(no red dotted area) + Green dotted area + CAM | 71.17 | 93.86 |
(f) Baseline + Local Branch1(no red dotted area) + Local Branch2(no red dotted area) + Green dotted area + CAM | 76.32 | 95.83 |
(g) Baseline + Local Branch1 + Local Branch2 + Green dotted area + CAM (Ours) | 77.12 | 96.30 |
