TY - GEN
T1 - Aircraft Detection in Remote Sensing Images Using YOLOX-DCSA
AU - Gao, Meijing
AU - Chen, Sibo
AU - Fan, Xiangrui
AU - Sun, Huanyu
AU - Chen, Xu
AU - Sun, Bingzhou
AU - Guan, Ning
N1 - Publisher Copyright:
© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2026.
PY - 2026
Y1 - 2026
N2 - This paper proposes an enhanced fine-grained target detection and recognition algorithm named YOLOX-DCSA specifically tailored for aircraft remote sensing images. Traditional fine-grained detection techniques struggle with small inter-category differences, challenging feature extraction, and low recognition accuracy in remote sensing images. To address these issues, YOLOX-DCSA integrates a DCSA attention module, which combines channel and spatial attention mechanisms with dilated convolution to expand the receptive field and improve discrimination among various aircraft categories. Additionally, depthwise separable convolution is employed in the feature pyramid network to reduce model parameters and enhance computational efficiency. A BD-CSP module is also designed to further enhance the receptive field and improve feature extraction capabilities for fine-grained targets. Experimental results demonstrate that YOLOX-DCSA outperforms existing mainstream target detection and recognition algorithms in terms of recognition accuracy, parameter size, and running time, thereby validating its effectiveness in identifying different types of aircraft targets in remote sensing images. The open-source code will be released at https://github.com/DEIRDRE1414/YOLOX-DCSA.git.
AB - This paper proposes an enhanced fine-grained target detection and recognition algorithm named YOLOX-DCSA specifically tailored for aircraft remote sensing images. Traditional fine-grained detection techniques struggle with small inter-category differences, challenging feature extraction, and low recognition accuracy in remote sensing images. To address these issues, YOLOX-DCSA integrates a DCSA attention module, which combines channel and spatial attention mechanisms with dilated convolution to expand the receptive field and improve discrimination among various aircraft categories. Additionally, depthwise separable convolution is employed in the feature pyramid network to reduce model parameters and enhance computational efficiency. A BD-CSP module is also designed to further enhance the receptive field and improve feature extraction capabilities for fine-grained targets. Experimental results demonstrate that YOLOX-DCSA outperforms existing mainstream target detection and recognition algorithms in terms of recognition accuracy, parameter size, and running time, thereby validating its effectiveness in identifying different types of aircraft targets in remote sensing images. The open-source code will be released at https://github.com/DEIRDRE1414/YOLOX-DCSA.git.
KW - Attention Mechanism
KW - Feature Pyramid
KW - Fine-grained Recognition
KW - Remote Sensing Image
KW - YOLOX-DCSA
UR - https://www.scopus.com/pages/publications/105038498631
U2 - 10.1007/978-981-95-6736-2_1
DO - 10.1007/978-981-95-6736-2_1
M3 - Conference contribution
AN - SCOPUS:105038498631
SN - 9789819567355
T3 - Communications in Computer and Information Science
SP - 3
EP - 16
BT - Advanced Computational Intelligence and Intelligent Informatics - 9th International Workshop, IWACIII 2025, Proceedings
A2 - Ma, Hongbin
A2 - Xin, Bin
A2 - She, Jinhua
A2 - Yoshida, Shinichi
PB - Springer Science and Business Media Deutschland GmbH
T2 - 9th International Workshop on Advanced Computational Intelligence and Intelligent Informatics, IWACIII 2025
Y2 - 31 October 2025 through 4 November 2025
ER -