Optical Remote Sensing Image Target Detection Based on Improved Feature Pyramid
Runxi Wei, Zhejun Feng, Zengyan Wu, Chaoran Yu, Baoming Song, Changqing C. Cao · IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing · 2023
At present, many deep convolution-based remote sensing image target detection methods have been developed and have achieved higher detection accuracy and faster detection rate. However, it does not perform well in the face of data sets with large target scale changes, multi-class and dense small targets. Therefore, solving the problem of scale change of remote sensing images is the focus of our research. An improved feature pyramid model named FE-FPN (Feature Enhancement Feature Pyramid Network) is presented in this paper. The FE-FPN utilizes a channel enhancement module (CEM), unpooling feature fusion (UPFF) and adaptive pooling spatial attention module (APSAM) to reduce information loss during the generation of feature maps as well as improve its capability to represent feature pyramids. The CEM is designed to expand the receptive field and learn important features adaptively, the UPFF is designed to improve the feature fusion mothed to avoid feature conflicts, and the APSAM is used to complement high-level feature information. The average precision of our models using ResNet50 is 2.0 % higher when FE-FPN is replaced by FPN in Cascade R-CNN. And the proposed FE-FPN model is quantitatively compared with several classical characteristic pyramid models, which proves that the performance of FE-FPN is superior to other models.