SFR-YOLO:改进YOLOv8的无人机图像小目标检测算法

doi:10.13474/j.cnki.11-2246.2025.0706

测绘通报 ›› 2025, Vol. 0 ›› Issue (7): 32-39.doi: 10.13474/j.cnki.11-2246.2025.0706

SFR-YOLO:改进YOLOv8的无人机图像小目标检测算法

孙基源¹, 纪松¹, 高定¹, 李凯², 张芮莹¹

1. 信息工程大学, 河南郑州 450001;
2. 军事科学院系统科学研究院, 北京 100044

收稿日期:2024-12-12 发布日期:2025-08-02
通讯作者: 纪松。E-mail:jisong_chxy@163.com
作者简介:孙基源(1995—),男,硕士生,主要从事遥感图像智能解译相关研究。E-mail:17836953603@139.com
基金资助:
国家自然科学基金(42371459);嵩山实验室项目(221100211000-4);河南省国家自然科学基金(222300420592)

SFR-YOLO: small target detection algorithm for UAV imagery based on improved YOLOv8

SUN Jiyuan¹, JI Song¹, GAO Ding¹, LI Kai², ZHANG Ruiying¹

1. Information Engineering University, Zhengzhou 450001, China;
2. Institute f System Engineering, Academy of Military Sciences, Beijing 100044, China

Received:2024-12-12 Published:2025-08-02

摘要/Abstract

摘要： 针对无人机图像中小目标像素占比少、特征易丢失,以及传统目标检测模型参数量大、难部署的问题,本文提出了一种基于改进YOLOv8的轻量化小目标检测算法SFR-YOLO。首先,设计了一个轻量化的共享细节增强检测头(SDCDH),它不仅通过共享卷积降低检测头参数量,还在共享层中引入细节增强卷积(DEConv),以增强细节特征表示能力;然后,使用增加浅层特征融合支路和删除深层卷积的加权双向特征金字塔网络(BIFPN),改进特征融合网络,以提高小目标检测性能;最后,设计了CRFA模块,结合空间注意力和感受野特征,提升了模型主干网络的特征提取能力。试验结果表明,SFR-YOLO在VisDrone2019数据集中,相对YOLOv8n算法mAP提升了3.8%,不仅小目标检测效果得以提升,而且满足了模型部署的要求;此外,通过SFR-YOLO在CARPK数据集上的迁移试验,进一步验证了本文方法的有效性。

关键词: 小目标检测, YOLOv8, 无人机图像, 轻量化

Abstract: Addressing the issues of small targets in drone imagery having a low pixel ratio,leading to easy loss of features,as well as the large parameter count and deployment challenges of traditional object detection models,this paper proposes a lightweight small object detection algorithm named SFR-YOLO based on an improved YOLOv8.Firstly,this paper introduces a lightweight Shared detail-enhanced convolution detection head (SDCDH),which not only reduces the number of parameters in the detection head by sharing convolutions but also enhances the representation of detailed features by introducing detail-enhanced convolution (DEConv) in the shared layers.Secondly,the feature fusion network is improved using a weighted bidirectional feature pyramid network (BIFPN) with added shallow feature fusion branches and the removal of deep convolution,which boosts the detection performance for small objects.Finally,this paper designs a CRFA module that combines spatial attention and receptive field features to enhance the feature extraction capability of the model's backbone network.Experimental results demonstrate that SFR-YOLO achieves a 3.8%improvement in mean average precision (mAP) compared to the YOLOv8n algorithm on the VisDrone2019 dataset,SFR-YOLO not only enhances the detection of small objects but also meets the requirements for model deployment.Additionally,transfer experiments of SFR-YOLO on the CARPK dataset further validating the effectiveness of the proposed method in this paper.

Key words: small target detection, YOLOv8, drone imagery, lightweight

中图分类号:

P237

孙基源, 纪松, 高定, 李凯, 张芮莹. SFR-YOLO:改进YOLOv8的无人机图像小目标检测算法[J]. 测绘通报, 2025, 0(7): 32-39.

SUN Jiyuan, JI Song, GAO Ding, LI Kai, ZHANG Ruiying. SFR-YOLO: small target detection algorithm for UAV imagery based on improved YOLOv8[J]. Bulletin of Surveying and Mapping, 2025, 0(7): 32-39.

参考文献

[1] 曹家乐,李亚利,孙汉卿,等.基于深度学习的视觉目标检测技术综述[J].中国图象图形学报,2022,27(6): 1697-1722.
[2] GIRSHICK R.Fast R-CNN[C]//Proceedings of 2015 IEEE International Conference on Computer Vision.Santiago,Chile:IEEE,2015:1440-1448.
[3] REN Shaoqing,HE Kaiming,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.
[4] HE Kaiming,GKIOXARI G,DOLLÁR P,et al.Mask R-CNN[C]//Proceedings of 2017 IEEE International Conference on Computer Vision.Venice:IEEE,2017:2980-2988.
[5] LIU Wei,ANGUELOV D,ERHAN D,et al.SSD:single shot MultiBox detector[EB/OL].[2024-12-01].https://arxiv.org/abs/1512.02325v5.
[6] REDMON J,DIVVALA S,GIRSHICK R,et al.You only look once:unified,real-time object detection[C]//Proceedings of 2016 IEEE Conference on Computer Vision and Pattern Recognition.Las Vegas,NV:IEEE,2016:779-788.
[7] ZHANG Zhengxin.Drone-YOLO:an efficient neural network method for target detection in drone images[J].Drones,2023,7(8):526.
[8] HUANGFU Zhongmin,LI Shuqing.Lightweight you only look once v8:an upgraded you only look once v8 algorithm for small object identification in unmanned aerial vehicle images[J].Applied Sciences,2023,13(22):12369.
[9] 曹利,徐慧英,谢刚,等.ASOD-YOLO:基于YOLOv8n改进的航空小目标检测算法[EB/OL].[2024-11-25].https://link.cnki.net/urlid/43.1258.tp.20240926.1558.002.
[10] 陈卫彪,贾小军,朱响斌,等.基于DSM-YOLOv5的无人机航拍图像目标检测[J].计算机工程与应用,2023,59(18):1-11.
[11] CAO Lijia,SONG Pinde,WANG Yongchao,et al.An improved lightweight real-time detection algorithm based on the edge computing platform for UAV images[J].Electronics,2023,12(10):2274.
[12] CHEN Zixuan,HE Zewei,LU Zheming.DEA-net:single image dehazing based on detail-enhanced convolution and content-guided attention[J].IEEE Transactions on Image Processing,2024,33:1002-1015.
[13] ZHANG Xin,LIU Chen,YANG Degang,et al.RFAConv:innovating spatial attention and standard convolutional operation[EB/OL].[2024-12-02].https://arxiv.org/abs/2304.03198v6.
[14] WU Y,HE K.Group normalization[C]//Proceedings of 2018 European Conference on Computer Vision.Berlin:Springer,2018:3-19.
[15] TAN Mingxing,PANG Ruoming,LE Q V.EfficientDet:scalable and efficient object detection[C]//Proceedings of 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition.Seattle,WA:IEEE,2020:10778-10787.
[16] DU Dawei,WEN Longyin,ZHU Pengfei,et al.VisDrone-DET2020:the vision meets drone object detection in image challenge results[C]//Proceedings of 2020 Computer Vision-ECCV 2020 Workshops.Cham:Springer International Publishing,2020:692-712.
[17] HSIEH M R,LIN Y L,HSU W H.Drone-based object counting by spatially regularized regional proposal network[C]//Proceedings of 2017 IEEE International Conference on Computer Vision.Venice:IEEE,2017:4165-4173.
[18] LIN T Y,GOYAL P,GIRSHICK R,et al.Focal loss for dense object detection[C]//Proceedings of 2017 IEEE International Conference on Computer Vision.Venice:IEEE,2017:2999-3007.
[19] ZHAO Yian,LV Wenyu,XU Shangliang,et al.DETRs beat YOLOs on real-time object detection[C]//Proceedings of 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition.Seattle,WA:IEEE,2024:16965-16974.
[20] GLENN J.YOLOv5[EB/OL].[2024-11-19].https://github.com/ultralytics/yolov5.
[21] WANG C Y,YEH I H,LIAO H M.YOLOv9:learning what you want to learn using programmable gradient information[EB/OL].[2024-11-19].https://arxiv.org/abs/2402.13616v2.
[22] WANG Ao,CHEN Hui,LIU Lihao,et al.YOLOv10:real-time end-to-end object detection[EB/OL].[2024-11-20].https://arxiv.org/abs/2405.14458v2.
[23] KHANAM R,HUSSAIN M.YOLOv11:an overview of the key architectural enhancements[EB/OL].[2024-11-20].https://arxiv.org/abs/2410.17725v1.

SFR-YOLO:改进YOLOv8的无人机图像小目标检测算法

SFR-YOLO: small target detection algorithm for UAV imagery based on improved YOLOv8

PDF

可视化

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 13

编辑推荐

Metrics

本文评价

[1]	席泽馨, 李佳仪, 谢昊, 甘文建, 周杨. GDS:无人机图像引导的跨视角图像地理定位[J]. 测绘通报, 2025, 0(7): 66-72.
[2]	蓝贵文, 徐梓睿, 任新月, 钟展, 郭瑞东, 范冬林. 基于YOLOv8n改进的航拍输电线路图像多类电力部件检测算法[J]. 测绘通报, 2024, 0(9): 38-43,49.
[3]	石一剑, 谭海, 钟旭辉. 改进YOLOv5的国产光学影像辐射异常检测方法[J]. 测绘通报, 2024, 0(8): 60-65,72.
[4]	姚楚羡, 蔡皓楠, 张远波, 唐可懿, 詹璐, 周宝定. 基于轻量化车载设备的道路病害检测方法[J]. 测绘通报, 2024, 0(5): 147-150.
[5]	贺彪, 唐骜巍, 蒯希, 徐海, 肖佳栋. 融合IFC语义信息与几何相似性的BIM构件实例信息提取方法[J]. 测绘通报, 2024, 0(5): 96-102.
[6]	李玉婷, 袁振超, 张丽. 基于改进YOLOv8的交通标识检测方法[J]. 测绘通报, 2024, 0(12): 106-110.
[7]	史与正, 陈梦华, 黄煜, 张彤蕴, 周宪. 实景三维模型的建筑物单体模型框架搭建[J]. 测绘通报, 2023, 0(6): 161-166.
[8]	张光伟, 吴昊, 郭震冬. 实景三维多源数据场景融合[J]. 测绘通报, 2022, 0(8): 155-159.
[9]	郭泽, 谭衢霖, 秦晓春, 贺佳, 胡吉平. 基于Cesium的铁路线路三维场景空间信息服务[J]. 测绘通报, 2022, 0(11): 144-148.
[10]	刘继, 杨军. 遥感影像目标检测的轻量化密集连接网络[J]. 测绘通报, 2022, 0(10): 37-43,55.
[11]	王文凯, 黄昆学, 李梦, 王业明. 倾斜摄影测量数据轻量化处理技术研究[J]. 测绘通报, 2020, 0(9): 123-126.
[12]	王玮. 基于三维GIS的铁路BIM空间信息系统构建及其工程应用[J]. 测绘通报, 2020, 0(7): 138-142.
[13]	王晓丽戴华阳余涛谢东海吴俣 . 基于多分辨率融合的无人机图像拼接匀色研究[J]. 测绘通报, 2013, 0(6): 27-30.