UAV Target Detection Algorithm with Improved Yolov5

doi:10.3778/j.issn.1002-8331.2304-0324

Abstract

Abstract: The characteristics of high density, small target and wide coverage of aerial images in UAV scenes make the existing target detectors prone to error detection and omission, and in order to improve the recognition accuracy, a target detection model with improved Yolov5 is proposed. Firstly, the feature extraction capability of the model is increased by adopting the gradient flow-rich C2F module. Secondly, the up-sampling operator CARAFE（content-aware reassembly of features） is introduced to increase the perceptual field for data feature fusion and improve the feature pyramid network performance. Finally, the model recognition accuracy is improved by adopting a global dynamic label assignment strategy. The validation by VisDrone2019 dataset shows that the improved model achieves an average accuracy mAP value of 65.3%, which is 24.7 percentage points higher than the traditional model, and can more accurately complete the target-specific detection task during aerial photography.

Key words: unmanned aerial vehicle（UAV）, Yolov5, CARAFE algorithm, OTA tag assignment strategy

摘要： 无人机场景下航拍图像存在密度高、目标小、覆盖范围广等特性，使得现有的目标检测器容易出现错检漏检的现象，为了提高识别的精度，提出了一种改进Yolov5的目标检测模型。通过采用梯度流丰富的C2F模块增加模型的特征提取能力。引入上采样算子CARAFE（content-aware reassembly of features）增加感受野进行数据特征融合，提升特征金字塔网络性能。通过采用全局性动态标签分配策略，提高模型识别准确率。通过VisDrone2019数据集验证表明，改进后的模型平均精度mAP值达到65.3%，较传统模型提升了24.7个百分点，可以更加准确地完成航拍过程中针对目标的检测任务。

关键词: 无人机, Yolov5, CARAFE算子, OTA标签分配策略

CHEN Fankai, LI Shixin. UAV Target Detection Algorithm with Improved Yolov5[J]. Computer Engineering and Applications, 2023, 59(18): 218-225.

陈范凯, 李士心. 改进Yolov5的无人机目标检测算法[J]. 计算机工程与应用, 2023, 59(18): 218-225.

References

[1] KATTENBORN T，LEITLOFF J，SCHIEFER F，et al.Review on convolutional neural networks（CNN） in vegetation remote sensing[J].ISPRS Journal of Photogrammetry and Remote Sensing，2021，173：24-49.
[2] CHENG B，WEI Y，SHI H，et al.Revisiting RCNN：on awakening the classification power of faster RCNN[C]//Proceedings of the European Conference on Computer Vision（ECCV），2018：453-468.
[3] WANG X，SHRIVASTAVA A，GUPTA A.A-fast-RCNN：hard positive generation via adversary for object detection[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition，2017：2606-2615.
[4] 李伟山，卫晨，王琳.改进的Faster RCNN煤矿井下行人检测算法[J].计算机工程与应用，2019，55（4）：200-207.
LI W H，WEI C，WANG L.lmproved faster RCNN approach for pedestrian detection in underground coal mine[J].Computer Engineering and Applications，2019，55（4）：200-207.
[5] WANG J，LUO J，LIU B，et al.Automated diabetic retinopathy grading and lesion detection based on the modified R‐FCN object‐detection algorithm[J].IET Computer Vision，2020，14（1）：1-8.
[6] CAO J，CHOLAKKAL H，ANWER R M，et al.D2det：towards high quality object detection and instance segmentation[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition，2020：11485-11494.
[7] 阳珊，王建，胡莉，等.改进RetinaNet的遮挡目标检测算法研究[J].计算机工程与应用，2022，58（11）：209-214.
YANG S，WANG J，HU L，et al.Research on occluded object detection by improved RetinaNet[J].Computer Engineering and Applications，2022，58（11）：209-214.
[8] 张灵灵，王鹏，李晓艳，等.基于优化SSD的低空无人机检测方法[J].计算机工程与应用，2022，58（16）：204-212.
ZHANG L L，WANG P，LI X Y，et al.Low-altitude UAV detection method based on optimized SSD[J].Computer Engineering and Applications，2022，58（16）：204-212.
[9] ZHAN W，SUN C，WANG M，et al.An improved Yolov5 real-time detection method for small objects captured by UAV[J].Soft Computing，2022，26：361-373.
[10] NEPAL U，ESLAMIAT H.Comparing YOLOv3，YOLOv4 and YOLOv5 for autonomous landing spot detection in faulty UAVs[J].Sensors，2022，22（2）：464.
[11] ZHAO J，ZHANG X，YAN J，et al.A wheat spike detection method in UAV images based on improved YOLOv5[J].Remote Sensing，2021，13（16）：3095.
[12] LIU W，QUIJANO K，CRAWFORD M M.YOLOv5-tassel：detecting tassels in RGB UAV imagery with improved YOLOv5 based on transfer learning[J].IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing，2022，15：8085-8094.
[13] 程江川，王伟，康林，等.基于改进YOLOv5与嵌入式平台的多旋翼无人机检测算法[J].兵工自动化，2023，42（4）：74-78.
CHENG J C，WANG W，KANG L，et al.Multi-rotor UAV detection algorithm based on improved YOLOv5 and embedded platform[J].Ordnance Industry Automation，2023，42（4）：74-78.
[14] 李利霞，王鑫，王军，等.基于特征融合与注意力机制的无人机图像小目标检测算法[J].图学学报，2023，44（4）：658-666.
LI L X，WANG X，WANG J，et al.Small object detection algorithm in UAV image based on feature fusion and attention mechanism[J].Journal of Graphics，2023，44（4）：658-666.
[15] 李杨，武连全，杨海涛，等.一种无人机视角下的小目标检测算法[J/OL].红外技术：1-7[2023-04-01].http：//kns.cnki.net/kcms/detail/53.1053.tn.20230316.1014.002.html.
LI Y，WU L Q，YANG H T，et al.A small target detection algorithm from UAV perspective[J/OL].Infrared Technology：1-7[2023-04-01].http：//kns.cnki.net/kcms/detail/53.1053.tn.20230316.1014.002.html.
[16] 甄然，刘颖，孟凡华，等.基于YOLOv5的轻量化目标检测算法[J].无线电工程，2023，53（6）：1242-1250.
ZHEN R，LIU Y，MENG F H，et al.Lightweight target detection algorithm based on YOLOv5[J].Radio Engineering，2023，53（6）：1242-1250.
[17] 薛珊，卢涛，吕琼莹，等.基于多尺度融合和轻量化网络的无人机目标检测算法[J/OL].湖南大学学报（自然科学版）：1-13[2023-04-01].http：//kns.cnki.net/kcms/detail/43.1061.N.20230320.1238.004.html.
XUE S，LU T，LV Q Y，et al.The Drone target detection algorithm based on multi-scale fusion andlightweight network[J/OL].Journal of Hunan University（Natural Sciences）：1-13[2023-04-01].http：//kns.cnki.net/kcms/detail/43.
1061.N.20230320.1238.004.html.
[18] 谢椿辉，吴金明，徐怀宇.改进YOLOv5的无人机影像小目标检测算法[J].计算机工程与应用，2023，59（9）：198-206.
XIE C H，WU J M，XU H Y.Small object detection algorithm based on improved YOLOv5 in UAV lmage[J].Computer Engineering and Applications，2023，59（9）：198-206.
[19] 韩佼志，王红雨，吴昌学，等.基于YOLOv5的单目视觉无人机检测与定位方法[J].飞行力学，2023，41（3）：61-66.
HAN J Z，WANG H Y，WU C X，et al.Monocular vision detection and localization method of UAV based on YOLOv5[J].Flight Dynawics，2023，41（3）：61-66.
[20] WANG J Q，CHEN K，XU R，et al.CARAFE：content-aware reassembly of features[C]//IEEE International Conference on Computer Vision，2019：3007-3016.
[21] GE Z，LIU S T，LI Z M，et al.OTA：optimal transport assignment for object detection[C]//2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition，2021：303-312.
[22] 张上，张岳，王恒涛，等.轻量化无人机遥感图像小目标检测算法[J/OL].无线电工程：1-9[2023-04-17].http：//kns.cnki.net/kcms/detail/13.1097.tn.20230414.1639.013.html.
ZHANG S，ZHANG Y，WANG H T，et al.Lightweight small target detection algorithm in UAV remote sensing image[J/OL].Radio Engineering：1-9[2023-04-17].http：//kns.cnki.net/kcms/detail/13.1097.tn.20230414.1639.013.html.
[23] CHEN J，KAO S，HE H，et al.Run，don’t walk：chasing higher FLOPS for faster neural networks[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition，2023：12021-12031.
[24] XU X，JIANG Y，CHEN W，et al.DAMO-YOLO：a report on real-time object detection design[J].arXiv：2211.15444，2022.
[25] LI H，LI J，WEI H，et al.Slim-neck by GSConv：a better design paradigm of detector architectures for autonomous vehicles[J].arXiv：2206.02424，2022.
[26] WANG C Y，BOCHKOVSKIY A，LIAN H Y M.YOLOv7：trainable bag-of-freebies sets new state-of-the-art for real-time object detectors[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition，2023：7464-7475.
[27] 刘展威，陈慈发，董方敏.基于YOLOv5s的航拍小目标检测改进算法研究[J/OL].无线电工程：1-10[2023-04-30].http：//kns.cnki.net/kcms/detail/13.1097.TN.20230411.1645.
026.html.
LIU Z W，CHEN C F，DONG F M.lmproved aerial small object detection algorithm based on YOLOv5s[J/OL].Radio Engineering：1-10[2023-04-30].http：//kns.cnki.net/kcms/detail/13.1097.TN.20230411.1645.026.html.