双向多尺度特征融合的无人机检测算法

doi:10.3778/j.issn.1002-8331.2407-0401

摘要/Abstract

摘要： 针对在公园、学校、机场等公共区域的复杂背景中，“黑飞”无人机目标尺度多变、模糊和遮挡等导致不易识别的问题，提出一种改进YOLOv8n的无人机目标检测算法。通过在C2f模块中融合RepViTBlock结构和高效多尺度注意力机制EMA（efficient multi-scale attention）改进Bottleneck块，设计了C2f-RVB模块，在减少参数的情况下增强模型对多尺度特征信息的提取能力；在颈部特征融合网络中构建动态边界融合模块DBFFPN（dynamic boundary fusion feature pyramid network），并新增小目标检测层，聚合浅层的边界信息和深层的语义信息，提高模型抗遮挡检测能力；在损失函数部分，提出MFShape-IoU替换原模型CIoU，使得模型更关注边界框自身形状和尺度信息，聚集困难样本，提高目标定位精度。在公开数据集CBD上进行实验，结果表明，改进后的算法相比YOLOv8n，mAP@0.5提升4.1个百分点达到93.0%，mAP@0.5：0.95提升4.2个百分点达到57.1%。同时算法精度高于YOLOv8s，复杂度远低于YOLOv8s，符合在移动设备上部署的需求。

关键词: 复杂背景, 反无人机, 小目标检测, YOLOv8, 注意力机制

Abstract: A UAV target detection algorithm with improved YOLOv8n is proposed to address the difficulty of recognizing “black-flying” UAV targets in complex environments such as parks, schools, and airports, where targets vary in scale, are blurred, and occluded. The C2f-RVB module is designed by fusing the RepViTBlock structure and efficient multi-scale attention (EMA) to improve the Bottleneck block in the C2f module, enhancing the multi-scale feature extraction capabilities of the model with fewer parameters. Furthermore, a dynamic boundary fusion module DBFFPN (dynamic boundary fusion feature pyramid network) is constructed in the neck feature fusion network, and a new small target detection layer is added to aggregate shallow boundary information and deep semantic information, thereby enhancing the ability of the model to detect obscuration. In the loss function section, MFShape-IoU is proposed as a replacement for the original model CIoU, encourages the model to focus on the shape and scale of bounding box, while aggregating challenging samples, ultimately improving target localization accuracy. Experiments are conducted on the public dataset CBD, and the results demonstrate that the enhanced algorithm exhibits improvements ranging from 4.1 percentage points to 93.0% on mAP@0.5, and 4.2 percentage points to 57.1% on mAP@0.5:0.95 in comparison to YOLOv8n. Meanwhile, the algorithm exhibits higher accuracy compared to YOLOv8s and significantly lower complexity, meets the needs of deployment on mobile devices.

Key words: complex background, anti-drone, small target detection, YOLOv8, attention mechanism

汤栎, 贾渊, 张玉宁. 双向多尺度特征融合的无人机检测算法[J]. 计算机工程与应用, 2025, 61(10): 267-278.

TANG Yue, JIA Yuan, ZHANG Yuning. UAV Detection Algorithm Based on Bidirectional Multi-Scale Feature Fusion[J]. Computer Engineering and Applications, 2025, 61(10): 267-278.

参考文献

[1] JIN X Q, ZHANG D W, WU Q E, et al. Improved SiamCAR with ranking-based pruning and optimization for efficient UAV tracking[J]. Image and Vision Computing, 2024, 141: 104886.
[2] HU Y Y, WU X J, ZHENG G D, et al. Object detection of UAV for anti-UAV based on improved YOLO v3[C]//Proceedings of the 2019 Chinese Control Conference. Piscataway: IEEE, 2019: 8386-8390.
[3] 张瑞鑫, 黎宁, 张夏夏, 等. 基于优化CenterNet的低空无人机检测方法[J]. 北京航空航天大学学报, 2022, 48(11): 2335-2344.
ZHANG R X, LI N, ZHANG X X, et al. Low-altitude UAV detection method based on optimized CenterNet[J]. Journal of Beijing University of Aeronautics and Astronautics, 2022, 48(11): 2335-2344.
[4] 包文歧, 谢立强, 徐才华, 等. 基于YOLOv5的微小型无人机实时探测方法[J]. 兵器装备工程学报, 2022, 43(5): 232-237.
BAO W Q, XIE L Q, XU C H, et al. Real-time detection method of micro UAV based on YOLOv5[J]. Journal of Ordnance Equipment Engineering, 2022, 43(5): 232-237.
[5] LIU B L, LUO H. An improved YOLOv5 for multi-rotor UAV detection[J]. Electronics, 2022, 11(15): 2330.
[6] 彭晏飞, 赵涛, 陈炎康, 等. 基于上下文信息与特征细化的无人机小目标检测算法[J]. 计算机工程与应用, 2024, 60(5): 183-190.
PENG Y F, ZHAO T, CHEN Y K, et al. UAV small object detection algorithm based on context information and feature refinement[J]. Computer Engineering and Applications, 2024, 60(5): 183-190.
[7] LV Y W, AI Z Q, CHEN M F, et al. High-resolution drone detection based on background difference and SAG-YOLOv5s[J]. Sensors, 2022, 22(15): 5825.
[8] 薛珊, 王亚博, 吕琼莹, 等. 基于YOLOX-drone的反无人机系统抗遮挡目标检测算法[J]. 工程科学学报, 2023, 45(9): 1539-1549.
XUE S, WANG Y B, LV Q Y, et al. Anti-occlusion target detection algorithm for anti-UAV system based on YOLOX-drone[J]. Chinese Journal of Engineering, 2023, 45(9): 1539-1549.
[9] GLENN J. YOLOv5 releases v6.1[CP/OL]. (2022-02-07)[2024-06-20]. https://github.com/ultralytics/yolov5/releases/tag/v6.1.
[10] WANG C Y, BOCHKOVSKIY A, LIAO H M. YOLOv7: trainable bag-of-freebies sets new state-of-the-art for real-time object detectors[C]//Proceedings of the 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2023: 7464-7475.
[11] 谢威宇, 张强. 基于深度学习的图像中无人机与飞鸟检测研究综述[J]. 计算机工程与应用, 2024, 60(8): 46-55.
XIE W Y, ZHANG Q. Review on detection of drones and birds in photoelectric images based on deep learning convolutional neural network[J]. Computer Engineering and Applications, 2024, 60(8): 46-55.
[12] WANG A, CHEN H, LIN Z J, et al. RepViT: revisiting mobile CNN from ViT perspective[C]//Proceedings of the 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2024: 15909-15920.
[13] DING X H, ZHANG X Y, MA N N, et al. RepVGG: making VGG-style ConvNets great again[C]//Proceedings of the 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2021: 13728-13737.
[14] HE J J, WANG Y C, WANG Y T, et al. A lightweight road crack detection algorithm based on improved YOLOv7 model[J]. Signal, Image and Video Processing, 2024, 18(1): 847-860.
[15] HU J, SHEN L, SUN G. Squeeze-and-excitation networks[C]//Proceedings of the 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2018: 7132-7141.
[16] OUYANG D L, HE S, ZHANG G Z, et al. Efficient multi-scale attention module with cross-spatial learning[C]//Proceedings of the 2023 IEEE International Conference on Acoustics, Speech and Signal Processing. Piscataway: IEEE, 2023: 1-5.
[17] TANG F L, XU Z X, HUANG Q M, et al. DuAT: dual-aggregation transformer network for medical image segmentation[C]//Proceedings of the 6th Chinese Conference on Pattern Recognition and Computer Vision. Singapore: Springer, 2023: 343-356.
[18] ZHANG Y F, REN W Q, ZHANG Z, et al. Focal and efficient IOU loss for accurate bounding box regression[J]. Neurocomputing, 2022, 506: 146-157.
[19] GEVORGYAN Z. SIoU loss: more powerful learning for bounding box regression[J]. arXiv:2205.12740, 2022.
[20] 周孟然, 王澳. 基于DETR的轻量级遥感图像目标检测算法[J/OL]. 重庆工商大学学报(自然科学版) [2024-06-25]. http://kns.cnki.net/kcms/detail/50.1155.N.20240328.1703. 004.html.
ZHOU M R, WANG A. Lightweight remote sensing image object detection algorithm based on DETR[J/OL]. Journal of Chongqing Technology and Business University (Natural Science Edition) [2024-06-25]. http://kns.cnki.net/kcms/detail/50.1155.N.20240328.1703.004.html.
[21] ZHANG H, ZHANG S. Shape-IOU: more accurate metric considering bounding box shape and scale[J]. arXiv:2312. 17663, 2023.
[22] TONG Z J, CHEN Y H, XU Z W, et al. Wise-IoU: bounding box regression loss with dynamic focusing mechanism[J]. arXiv:2301.10051, 2023.
[23] MUNIR A, SIDDIQUI A J, ANWAR S. Investigation of UAV detection in images with complex backgrounds and rainy artifacts[C]//Proceedings of the 2024 IEEE/CVF Winter Conference on Applications of Computer Vision Workshops. Piscataway: IEEE, 2024: 232-241.
[24] ZHENG Y, CHEN Z, LV D L, et al. Air-to-air visual detection of micro-UAVs: an experimental evaluation of deep learning[J]. IEEE Robotics and Automation Letters, 2021, 6(2): 1020-1027.
[25] COLUCCIA A, FASCISTA A, SCHUMANN A, et al. Drone vs. bird detection: deep learning algorithms and results from a grand challenge[J]. Sensors, 2021, 21(8): 2824.
[26] ZHAO J, ZHANG J S, LI D D, et al. Vision-based anti-UAV detection and tracking[J]. IEEE Transactions on Intelligent Transportation Systems, 2022, 23(12): 25323-25334.
[27] YANG L X, ZHANG R Y, LI L D, et al. SimAM: a simple, parameter-free attention module for convolutional neural net-works[C]//Proceedings of the 38th International Conference on Machine Learning, 2021: 11863-11874.
[28] WOO S, PARK J, LEE J Y, et al. CBAM: convolutional block attention module[C]//Proceedings of the 15th European Conference on Computer Vision. Cham: Springer, 2018: 3-19.
[29] XU W, WAN Y. ELA: efficient local attention for deep convolutional neural networks[J]. arXiv:2403.01123, 2024.
[30] REZATOFIGHI H, TSOI N, GWAK J, et al. Generalized intersection over union: a metric and a loss for bounding box regression[C]//Proceedings of the 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2019: 658-666.
[31] SELVARAJU R R, COGSWELL M, DAS A, et al. Grad-CAM: visual explanations from deep networks via gradient-based localization[J]. International Journal of Computer Vision, 2020, 128(2): 336-359.
[32] REDMON J, FARHADI A. YOLOv3: an incremental improvement[J]. arXiv:1804.02767, 2018.
[33] WANG A, CHEN H, LIU L H, et al. YOLOv10: real-time end-to-end object detection[J]. arXiv:2405.14458, 2024.
[34] REN S Q, HE K M, GIRSHICK R, et al. Faster R-CNN: towards real-time object detection with region proposal networks[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 39(6): 1137-1149.
[35] HE K M, GKIOXARI G, DOLLáR P, et al. Mask R-CNN[C]//Proceedings of the 2017 IEEE International Conference on Computer Vision. Piscataway: IEEE, 2017: 2980-2988.
[36] LIN T Y, GOYAL P, GIRSHICK R, et al. Focal loss for dense object detection[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2020, 42(2): 318-327.
[37] ZHANG H, LI F, LIU S L, et al. DINO: DETR with improved denoising anchor boxes for end-to-end object detection[J]. arXiv:2203.03605, 2022.
[38] FENG C J, ZHONG Y J, GAO Y, et al. TOOD: task-aligned one-stage object detection[C]//Proceedings of the 2021 IEEE/CVF International Conference on Computer Vision. Piscataway: IEEE, 2021: 3490-3499.
[39] DU D W, ZHU P F, WEN L Y, et al. VisDrone-SOT2019: the vision meets drone single object tracking challenge results[C]//Proceedings of the 2019 IEEE/CVF International Conference on Computer Vision Workshops. Piscataway: IEEE, 2019: 199-212.
[40] 高民, 陈高华, 古佳欣, 等. FLM-YOLOv8: 一种轻量级的口罩佩戴检测算法[J]. 计算机工程与应用, 2024, 60(17): 203-215.
GAO M, CHEN G H, GU J X, et al. FLM-YOLOv8: lightweight mask wearing detection algorithm[J]. Computer Engineering and Applications, 2024, 60(17): 203-215.