尽管目标检测技术已日趋成熟,但针对遥感图像的目标检测仍面临诸多挑战,主要源于遥感图像中目标特征不明显、背景复杂、小目标占比高以及目标类别不均衡等问题,导致检测精度不足.本文提出了一种遥感图像目标检测算法(Remote Sensing-YOLOv8,RS-YOLOv8),重点优化了瓶颈层模块,通过提出多尺度特征自注意融合结构提升了对小尺度目标和多层级特征的敏感度与筛选能力,降低了漏检率.设计EMASlideLoss优化损失函数,增强了对复杂样本的处理能力,并引入DySample动态上采样策略提升上采样质量,并在一定程度上改善了因新增模块导致的模型参数量增加.在SIMD数据集上的实验表明,RS-YOLOv8显著改善了小目标的漏检情况,mAP达到83.7%,较YOLOv8提升3.7%,展现出优异的检测性能和广泛的应用前景.
Despite the significant advancements in object detection technology, object detection in remote sensing images still faces numerous challenges. These challenges mainly arise from the inconspicuous nature of objects, complex backgrounds, a high proportion of small objects, and imbalanced object classes in remote sensing images, leading to insufficient detection accuracy.This article proposes a remote sensing image target detection algorithm(Remote Sensing-YOLOv8, RS-YOLOv8). The focus is on optimizing the bottleneck layer module, and by proposing a multi-scale feature self-attention fusion structure, the sensitivity and screening capabilities for small-scale targets and multi-level features are improved, and the missed detection rate is reduced. The EMASlideLoss optimization loss function is designed to enhance the processing capabilities of complex samples, and the DySample dynamic upsampling strategy is introduced to improve the upsampling quality, and to a certain extent, improve the increase in the number of model parameters caused by the new modules. Experiments on the SIMD data set show that RS-YOLOv8 significantly improves the missed detection of small targets, with mAP reaching 83.7%, 3.7% higher than YOLOv8, showing excellent detection performance and broad application prospects.- DOI:10.3969/j.issn.1000-1565.2025.03.009基于改进YOLOv8的遥感图像目标检测柯淇晟1,花强1,张博2( 1. 河北大学 数学与信息科学学院,河北 保定 071002;2. 河北农业大学 理学院,河北 保定 071001)摘 要:尽管目标检测技术已日趋成熟,但针对遥感图像的目标检测仍面临诸多挑战,主要源于遥感图像中目标特征不明显、背景复杂、小目标占比高以及目标类别不均衡等问题,导致检测精度不足.本文提出了一种遥感图像目标检测算法(Remote Sensing-YOLOv8,RS-YOLOv8),重点优化了瓶颈层模块,通过提出多尺度特征自注意融合结构提升了对小尺度目标和多层级特征的敏感度与筛选能力,降低了漏检率.设计EMASlideLoss优化损失函数,增强了对复杂样本的处理能力,并引入DySample动态上采样策略提升上采样质量,并在一定程度上改善了因新增模块导致的模型参数量增加.在SIMD数据集上的实验表明,RS-YOLOv8显著改善了小目标的漏检情况,mAP达到83.7%,较YOLOv8提升3.7%,展现出优异的检测性能和广泛的应用前景.关键词:目标检测;YOLOv8;注意力模块;遥感图像中图分类号:TP391.7 文献标志码:A 文章编号:1000-1565(2025)03-0309-08Remote sensing image object detection based on improved YOLOv8KE Qisheng1, HUA Qiang1, ZHANG Bo2(1. School of Mathematics and Information Science, Hebei University, Baoding 07101,China) 2. School of Science, Hebei Agricultural University, Baoding 071001, China)Abstract: Despite the significant advancements in object detection technology, object detection in remote sensing images still faces numerous challenges. These challenges mainly arise from the inconspicuous nature of objects, complex backgrounds, a high proportion of small objects, and imbalanced object classes in remote sensing images, leading to insufficient detection accuracy.This article proposes a remote sensing image target detection algorithm(Remote Sensing-YOLOv8, RS-YOLOv8). The focus is on optimizing the bottleneck layer module, and by proposing a multi-scale feature self-attention fusion structure, the sensitivity and screening capabilities for small-scale targets and multi-level features are improved, and the missed detection rate is reduced. The EMASlideLoss optimization loss function is designed to enhance the processing capabilities of complex samples, and the DySample dynamic upsampling strategy is introduced to improve the upsampling quality, and to a certain extent, improve the increase in the number of model parameters caused by the new modules. Experiments on the SIMD data set show that RS-YOLOv8 significantly improves the missed detection of small targets, with mAP reaching 83.7%, 3.7% higher than YOLOv8, showing excellent detection performance and broad application prospects.- 收稿日期:2024-10-15;修回日期:2024-11-20 基金项目:科技部重点研发计划项目(2022YFE0196100) 第一作者:柯淇晟(2000—),男,河北大学在读硕士研究生,主要从事机器学习、计算机视觉方面研究.E-mail:1223482107@qq.com 通信作者:花强(1973—),男,河北大学教授,主要从事机器学习、人工神经网络方面的研究.E-mail:huaq@hbu.edu.cn 第3期柯淇晟等:基于改进YOLOv8的遥感图像目标检测河北大学学报(自然科学版) 第45卷
[1] 赵文清,康怿瑾,赵振兵,等.改进 YOLOv5s 的遥感图像目标检测[J]. 智能系统学报, 2022, 18(1): 86-95.DOI:10.11992/tis.202203013.
[2] 徐丹青,吴一全.光学遥感图像目标检测的深度学习算法研究进展[J].遥感学报,2024:1-30. DOI: 10.11834/jrs.20243166.
[3] ZHANG Z M, WARRELL J, TORR P H S. Proposal generation for object detection using cascaded ranking SVMs[C] //CVPR, Colorado Springs, CO, USA, IEEE, 2011: 1497-1504. DOI: 10.1109/CVPR.2011.5995411.
[4] SHI Z, YU X, JIANG Z,et al.Ship Detection in High-Resolution Optical Imagery Based on Anomaly Detector and Local Shape Feature[J].IEEETransactions on Geoscience and Remote Sensing, 2014, 52(8):4511-4523.DOI:10.1109/TGRS.2013.2282355.
[5] ZHU X K, LYU S C, WANG X, et al. TPH-YOLOv5: improved YOLOv5 based on transformer prediction head for object detection on drone-captured scenarios[C] //2021 IEEE/CVF International Conference on Computer Vision Workshops(ICCVW), Montreal, BC, Canada, IEEE, 2021: 2778-2788. DOI: 10.1109/ICCVW54120.2021.00312.
[6] XU X K, FENG Z J, CAO C Q, et al, An Improved Swin Transformer-Based Model for Remote Sensing Object Detection and Instance Segmentation[J].Remote Sensing,2021,13(23):4779.DOI:10.3390/rs13234779.
[7] LIU Z, LIN Y T, CAO Y, et al. Swin Transformer: hierarchical Vision Transformer using Shifted Windows[C] // 2021 IEEE/CVF International Conference on Computer Vision(ICCV), Montreal, QC, Canada, IEEE, 2021: 9992-10002. DOI: 10.1109/ICCV48922.2021.00986.
[8] WANG G, CHEN Y F, AN P, et al. UAV-YOLOv8: a small-object-detection model based on improved YOLOv8 for UAV aerial photography scenarios[J]. Sensors, 2023, 23(16): 7190. DOI: 10.3390/s23167190.
[9] 许德刚,王再庆,邢奎杰,等.改进YOLOv6的遥感图像目标检测算法[J].计算机工程与应用, 2024, 60(3): 119-128. DOI: 10.3778/j.issn.1002-8331.2304-0270.
[10] JIANG L B, LI X J. An efficient and accurate object detection algorithm and its application[C] //2020 IEEE 5th Information Technology and Mechatronics Engineering Conference(ITOEC). Chongqing, China, IEEE, 2020: 656-661. DOI: 10.1109/ITOEC49072.2020.9141752.
[11] BOCHKOVSKIY A, WANG C Y, LIAO H Y M. YOLOv4: optimal speed and accuracy of object detection[EB/OL]. 2020: arXiv: 2004.10934. http://arxiv.org/abs/2004.10934.
[12] LI C Y, LI L L, JIANG H L, et al. YOLOv6: a single-stage object detection framework for industrial applications[EB/OL]. 2022: arXiv: 2209.02976. http://arxiv.org/abs/2209.02976.
[13] WANG C Y, BOCHKOVSKIY A, LIAO H Y M. YOLOv7: trainable bag-of-freebies sets new state-of-the-art for real-time object detectors[C] //2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition(CVPR). Vancouver, BC, Canada. IEEE, 2023: 7464-7475. DOI: 10.1109/CVPR52729.2023.00721.
[14] WOO S, PARKJ, LEE J Y, et al. Cbam: Convolutional block attention module[C] //Proceedings of the European Conference on Computer Vision(ECCV), 2018:3-19. DOI: 10.1007/978-3-030-01234-2_1.
[15] YU Z P, HUANG H B, CHEN W J, et al. YOLO-FaceV2: a scale and occlusion aware face detector[J]. Pattern Recognit, 2024, 155: 110714. DOI: 10.1016/j.patcog.2024.110714.
[16] WANG J Q, CHEN K, XU R, et al. CARAFE: content-aware ReAssembly of FEatures[C] //2019 IEEE/CVF International Conference on Computer Vision(ICCV), Seoul, Korea(South), IEEE, 2019: 3007-3016. DOI: 10.1109/ICCV.2019.00310.
[17] LU H, LIU W Z, FU H T, et al. FADE: a task-agnostic upsampling operator for encoder–decoder architectures[J]. Int J Comput Vis, 2025,133: 151-172. DOI: 10.1007/s11263-024-02191-8.
[18] LU H, LIU W Z, YE Z X, et al. SAPA: similarity-aware point affiliation for feature upsampling[EB/OL]. 2022: arXiv: 2209.12866. http://arxiv.org/abs/2209.12866.
[19] XIA G S, BAI X, DING J, et al. DOTA: a large-scale dataset for object detection in aerial images[C] //2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, Salt Lake City, UT, USA, IEEE, 2018: 3974-3983. DOI: 10.1109/CVPR.2018.00418.
[20] HAROON M, SHAHZAD M, FRAZ M M. Multisized object detection using spaceborne optical imagery[J]. IEEE J Sel Top Appl Earth Obs Remote Sens, 2020, 13: 3032-3046. DOI: 10.1109/JSTARS.2020.3000317.
[21] TAN M X, PANG R M, LE Q V. EfficientDet: scalable and efficient object detection[C] //2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition(CVPR), Seattle, WA, USA, IEEE, 2020: 10778-10787. DOI: 10.1109/CVPR42600.2020.01079.
[22] CHEN Y F, ZHANG C Y, CHEN B, et al. Accurate leukocyte detection based on deformable-DETR and multi-level feature fusion for aiding diagnosis of blood diseases[J]. Comput Biol Med, 2024, 170: 107917. DOI: 10.1016/j.compbiomed.2024.107917.
[23] ZHAO G M, GE W F, YU Y Z. GraphFPN: graph feature pyramid network for object detection[C] //2021 IEEE/CVF International Conference on Computer Vision(ICCV), Montreal, QC, Canada, IEEE, 2021: 2743-2752. DOI: 10.1109/ICCV48922.2021.00276.
[24] WANG C C, HE W, NIE Y, et al. Gold-YOLO: efficient object detector via gather-and-distribute mechanism[EB/OL]. 2023: arXiv: 2309.11331. https://doi.org/10.48550/arXiv.2309.11331. (
