Small object detection algorithm is always a difficult research topic in remote sensing images. A remote sensing image object detection algorithm which is based on receptive field enhancement is proposed in this paper. Firstly, the backbone network is the SSD net. In order to form the feature map pyramid, the receptive field enhanced convolution module is added in feature extraction. Then, the feature maps of the low-level and the high-level are spliced, so he output feature map retains the rich context information. Because it contains the information in the low-level feature map and the high-level feature map, which are location information and semantic information, respectively. Finally, based on the idea of dilated convolution, the receptive field enhancement module is designed to make the receptive field of the feature map larger. And the feature information is filtered to reduce the aliasing effect when the feature map is spliced. After the experimental result on remote sensing image dataset NWPU vhr-10, the average detection accuracy of the algorithm is 90%, which improves the object detection accuracy, especially the detection accuracy of the small objects.
Remote Sensing Image Object Detection Algorithm Based on Receptive Field Enhancement
Lect. Notes Electrical Eng.
International Conference on Autonomous Unmanned Systems ; 2021 ; Changsha, China September 24, 2021 - September 26, 2021
Proceedings of 2021 International Conference on Autonomous Unmanned Systems (ICAUS 2021) ; Kapitel : 231 ; 2336-2343
2022-03-18
8 pages
Aufsatz/Kapitel (Buch)
Elektronische Ressource
Englisch
Hybrid receptive field network for small object detection on drone view
Elsevier | 2024
|Remote sensing image enhancement via bilateral filtering
IEEE | 2017
|