Research on Intelligent Recognition and Positioning Method of Civil Airport Bird Repellent Cannons Based on Improved YOLOV10

DOI: https://doi.org/10.62517/jbdc.202501415

Author(s)

Zheng Wang*, Xinlei Zhang, Chou Hu

Affiliation(s)

Qingdao Campus of Naval Aviation University, Qingdao, Shandong, China *Corresponding Author

Abstract

Addressing the challenges of bird control applications at civil airports—namely the difficulty of detecting distant small targets, strong background interference, and high real-time requirements—this paper proposes a YOLOv10n-attention-based recognition and localization method for bird control cannons. This approach utilizes YOLOv10n as its backbone. Without altering the pyramid and neck topology, it introduces global attention at the high-level branches to enhance response to weakly textured targets. The P5 branch employs a C2fCIB bottleneck structure to stabilize semantic representation across large receptive fields. The detection head retains the decoupled architecture with a DFL + IoU + CE loss combination, performing only “bird/non-bird” binary classification to meet engineering closed-loop requirements. Experiments on the self-built Airport-Birds dataset (airport surveillance frames with single-class annotations) demonstrate that our model achieves approximately mAP@0.50=0.83 at 640×640 input resolution with real-time performance around 30 FPS. This represents a stable improvement over the baseline without significantly increasing parameters or computational cost. Ablation studies further validate the synergistic contribution of attention mechanisms and the P5 bottleneck architecture to accuracy, recall, and localization robustness. By integrating recognition results with servo control mapping, the system completes the “detection-localization-lock-deterrence” engineering loop, meeting airport scenarios' demands for rapid response and high reliability. This work provides a reusable, deployable visual front-end solution for intelligent airport bird deterrence systems, laying the foundation for future multimodal fusion and edge deployment.

Keywords

Airport Bird Deterrence; Object Detection; YOLOv10; Attention Mechanism; Bird Deterrent Cannon

References

[1]Chen W, Liu J, Wu D, et al. A review of wide-area bird monitoring and early warning technologies. Journal of Beijing University of Aeronautics and Astronautics, 2025: 1-22. DOI: 10.13700/j.bh.1001-5965.2025.0427. [2]Du J, Shi F Y, Zhang Y B, et al. Bird diversity and bird strike prevention measures at airports in spring. China Science and Technology Information, 2023(15): 35-38. DOI: CNKI: SUN: XXJK.0.2023-15-012. [3]Zhu W M, Tu C B. Analysis and prevention measures of bird damage to power transmission lines. Science & Technology Vision, 2015(31): 267-268+271. DOI: 10.19694/j.cnki.issn2095-2457.2015.31.207. [4]Dai H F, Ming G Z, Zhao Y, et al. Introduction performance and cultivation techniques of ‘Shannong Su Pear’ in Feicheng, Shandong Province. Journal of Fruit Resources, 2022, 3(03): 91-93. DOI: 10.16010/j.cnki.14-1127/s.2022.03.028. [5]Wang C. Research on the application of linkage control in intelligent bird perception and repelling. Yancheng Institute of Technology, 2024. DOI: 10.44381/d.cnki.gycit.2024.000091. DOI: 10.44381/d.cnki.gycit.2024.000091. [6]Zhang H, Gong Y, Wang A J, et al. Research on bird damage identification methods for power transmission and transformation equipment based on improved YOLOv5. Journal of Nanjing Institute of Technology (Natural Science Edition), 2024, 22(02): 76-82. DOI: 10.13960/j.issn.1672-2558.2024.02.012. [7]Zhang X W. Research on airport bird detection and tracking technology based on image processing. Civil Aviation Flight University of China, 2025. DOI: 10.27722/d.cnki.gzgmh.2025.000076. [8]Ran X Y. Research on intrusion detection of transmission lines based on multi-source data fusion. Guangdong University of Technology, 2025. DOI: 10.27029/d.cnki.ggdgu.2025.001892. [9]Wang Q Y, Yao G Q, Fang C Y. Research on lightweight bird target detection and recognition model based on YOLOv8n in Poyang Lake. Journal of Jiangxi Normal University (Natural Science Edition), 2025, 49(01): 86-94. DOI: 10.16357/j.cnki.issn1000-5862.2025.01.11. [10]Han T. Research on bird target detection and recognition methods based on YOLOv8. Dalian Jiaotong University, 2025. DOI: 10.26990/d.cnki.gsltc.2025.001020. [11]Shi Y L. Research on airport bird identification and counting based on deep neural networks. Civil Aviation University of China, 2023. DOI: 10.27627/d.cnki.gzmhy.2023.000158. [12]Teng Y. Bird repelling recognition technology at airports based on multi-source bird feature fusion. Yancheng Institute of Technology, 2024. DOI: 10.44381/d.cnki.gycit.2024.000241. [13]Xiao J, He X Z, Cheng H L, et al. Aerial small object detection algorithm based on multi-scale feature enhancement. Journal of Zhejiang University (Engineering Edition), 2025: 1-13. Available from: https://link.cnki.net/urlid/33.1245.T.20251009.1118.006. [14]Zhang H J, Zheng P, Qiao W W, et al. A worm gear surface defect detection method based on improved YOLOv8n. Machine Tool & Hydraulics, 2025, 53(16): 10-17. DOI: CNKI: SUN: JCYY.0.2025-16-002. [15]Hou Y, Wu Y, Kou X R, et al. A small-object detection algorithm for UAV aerial images based on improved YOLOv8. Computer Engineering and Applications, 2025, 61(11): 83-92. [16]Xie R J. Implementation of real-time detection system based on lightweight YOLO. Information Systems Engineering, 2025(08): 12-15. DOI: CNKI: SUN: XXXT.0.2025-08-003. [17]Zhang J L, Gao C Q, Zhao B. YOLOv8 reservoir type identification method based on attention enhancement and re-parameterized bidirectional feature pyramid network. Progress in Geophysics, 2025: 1-17. Available from: https://link.cnki.net/urlid/11.2982.P.20251015.0945.034. [18]Zheng Y H, Chen E J, Wu F, et al. Small foreign object detection method for cigarette cutting production line based on improved YOLOv5s. Journal of Hefei University of Technology (Natural Science Edition), 2025, 48(09): 1183-1191. DOI: CNKI: SUN: HEFE.0.2025-09-005.