Acta Photonica Sinica, Volume. 53, Issue 4, 0415002(2024)
Enhancing Aircraft Object Detection in Complex Airport Scenes Using Deep Transfer Learning
Figures & Tables(7)
Fig. 2. Pre-trained and transfer-learned ResNet-50 model combined with the adjustment module
Table 1. Data for different categories in Aeroplane dataset.
View tableView in ArticleTable 1. Data for different categories in Aeroplane dataset.
Category Original image Labeled image Count Single aircraft 
900 Overlapped aircraft 
400 Small target 
400 Others 
300
Table 2. Experimental results of our method and other models on the Aeroplane dataset
View tableView in ArticleTable 2. Experimental results of our method and other models on the Aeroplane dataset
Category RetinaNet AP/% Inception-V3+FPN AP/% ResNet-34+FPN AP/% ResNet-50+FPN AP/% Ours AP/% Single aircraft （test 180） 85.8 81.4 83.6 84.5 89.4 Overlapped aircraft （test 80） 67.1 58.8 59.3 61.7 65.7 Small target （test 80） 80.5 72.6 76.9 78.2 82.6 All 81.3 74.0 78.6 80.0 82.2
Tools
Get Citation
Copy Citation Text
Dan ZHONG, Tiehu LI, Cheng LI. Enhancing Aircraft Object Detection in Complex Airport Scenes Using Deep Transfer Learning[J]. Acta Photonica Sinica, 2024, 53(4): 0415002
Paper Information
Category: Machine Vision
Received: Jan. 19, 2024
Accepted: Mar. 11, 2024
Published Online: May. 15, 2024
The Author Email: Dan ZHONG (henryzhongdan@mail.nwpu.edu.cn)
© Copyright 2018-2021 | Chinese Laser Press.
All Rights Reserved 沪ICP备15018463号-20