Acta Photonica Sinica, Volume. 49, Issue 7, 710004(2020)
Ship Detection Method in Remote Sensing Image Based on Feature Fusion
Figures & Tables(9)
Fig. 3. MSSDN overall framework and the feature fusion mode map of layer Conv8_2
Fig. 4. Comparison of feature maps before and after feature fusion
Fig. 6. Histograms of the distribution of the area size and aspect ratio of the bounding boxes
Fig. 8. Example of detection results of fuzzy ship targets by MSSDN and SSD
Table 1. Comparison of experimental results
View tableView in ArticleTable 1. Comparison of experimental results
Method Backbone Inputsize AP(%)@0.5 AP(%)@0.7 Speed/FPS FastR-CNN[ 11 ]VGG16 1 000×600 65.6 48.2 0.07 FasterR-CNN[ 12 ]VGG16 1 000×600 73.1 53.2 4.4 YOLO[ 13 ]GoogleNet 448×448 68.1 50.1 70.2 YOLOV2[ 14 ]DarkNet19 544×544 74.2 64.6 59.2 SSD[ 16 ]VGG16 300×300 76.4 68.7 63.6 DSSD[ 17 ]ResNet101 321×321 78.1 70.4 9.5 FSSD[ 18 ]VGG16 300×300 80.8 71.6 47.1 MSSDN(ours) VGG16 300×300 81.7 72.2 43.2
Tools
Get Citation
Copy Citation Text
Wen-xu SHI, Jin-hong JIANG, Sheng-li BAO. Ship Detection Method in Remote Sensing Image Based on Feature Fusion[J]. Acta Photonica Sinica, 2020, 49(7): 710004
Paper Information
Category: Image Processing
Received: Oct. 28, 2019
Accepted: --
Published Online: Aug. 25, 2020
The Author Email:
© Copyright 2018-2021 | Chinese Laser Press.
All Rights Reserved 沪ICP备15018463号-20