Fast Airplane Detection Based on Multi-Layer Feature Fusion of Fully Convolutional Networks

Table 1. Bounding box sizes of nine candidate regions
View table
Table 1. Bounding box sizes of nine candidate regions
Set 30×30 60×60 110×110 30×30 60×60 110×110 30×30 60×60 110×110
Proportion 1∶1 1∶1 1∶1 4∶5 4∶5 4∶5 5∶4 5∶4 5∶4
Size 30×30 60×60 110×110 27×33 53×67 98×122 33×27 67×53 122×98

Table 2. Simulation results of feature maps by fusing different layers
View table
Table 2. Simulation results of feature maps by fusing different layers
Layer 5 1,2,3 1,3,5 3,4,5
Precision /% 78.7 88.4 94.6 91.8
Recall /% 74.5 85.3 89.5 87.6

Table 3. Comparison of detection results of proposed method and four typical methods
View table
Table 3. Comparison of detection results of proposed method and four typical methods
Method Location-DBN BING-CNN Fast RCNN Faster RCNN Proposed method
Detection rate /% 84 86 80 79 93
False alarm rate /% 36 19 22 23 8
Average time /s >100 6.1 2.9 0.2 0.3

Tools

Get Citation

Copy Citation Text

Xin Peng, Xu Yuelei, Tang Hong, Ma Shiping, Li Shuai, Lü Chao. Fast Airplane Detection Based on Multi-Layer Feature Fusion of Fully Convolutional Networks[J]. Acta Optica Sinica, 2018, 38(3): 315003

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites