Table 6 Comparison of other methods combined with DenseNet on PASCAL VOC
From: Multiclass objects detection algorithm using DarkNet-53 and DenseNet for intelligent vehicles
Method | Data | Backbone | AP (%) | mAP (%) | ||||
|---|---|---|---|---|---|---|---|---|
Person | Bicycle | Bus | Car | Motorbike | ||||
Dsod [26] | 07 + + 12 | DS/64-192-48-1 | 84.6 | 85.3 | 83.6 | 80.6 | 86.8 | 84.18 |
DF-SSD [27] | 07 + + 12 | DenseNet-S-32-1 | 85.7 | 85.6 | 82.9 | 79.9 | 86.4 | 84.10 |
DS–YOLO [28] | 07 + + 12 | DarkNet-53 | 80.9 | 78.96 | 76.3 | 85.58 | 79.9 | 80.33 |
D-MIF [29] | 07 + | DenseNet-Evo | 85.12 | 85.65 | 86.10 | 89.50 | 82.86 | 85.85 |
ADFPNet [30] | 07 + + 12 | VGG-16 | 88.70 | 88.60 | 86.60 | 87.40 | 89.70 | 88.20 |
Ours | 07 + + 12 | DarkNet53 | 91.34 | 91.56 | 89.92 | 88.54 | 82.88 | 88.84 |