|本期目录/Table of Contents|

[1]张世豪,董峦,赵昀杰.基于YOLOX的小麦穗旋转目标检测[J].江苏农业科学,2024,52(20):157-164.
　Zhang Shihao,et al.Rotating target detection of wheat ears based on YOLOX[J].Jiangsu Agricultural Sciences,2024,52(20):157-164.
点击复制

基于YOLOX的小麦穗旋转目标检测(PDF)
分享到：

《江苏农业科学》[ISSN:1002-1302/CN:32-1214/S]

卷:
第52卷
期数:
2024年第20期
页码:
157-164
栏目:
果实智能检测
出版日期:
2024-10-20

文章信息/Info

Title:
Rotating target detection of wheat ears based on YOLOX
作者:
张世豪;  董峦;  赵昀杰
新疆农业大学计算机与信息工程学院/智能农业教育部工程研究中心/新疆农业信息化工程技术研究中心，新疆乌鲁木齐 830052
Author(s):
Zhang Shihao; et al
关键词:
目标检测; 小麦穗; 旋转矩形框; YOLOX; 坐标注意力模块; KL额度; 损失函数
Keywords:
-
分类号:
S126；TP391.41
DOI:
-
文献标志码:
A
摘要:
小麦穗检测对于农业估产和育种研究具有重要意义，但由于小麦穗角度和姿态多变且存在遮挡和尺度变化等因素，给目标检测带来较大困难，提出一种针对小麦穗旋转目标检测的改进方法YOLOX-RoC，该方法在YOLOX基础上使用旋转矩形框代替水平矩形框，更好地拟合小麦穗的轮廓和方向，减少背景干扰和重叠区域，使模型更具灵活性，更准确地捕捉小麦穗的特征；添加坐标注意力模块并采用KL散度损失函数代替交叉熵损失函数，提高对旋转目标的感知能力并解决旋转敏感度的误差度量问题，优化旋转目标的定位精度。利用基于图像合成的 Copy-Paste 数据增强方法，生成更多的训练样本以提高模型对不同尺度、姿态和遮挡情况的泛化能力，提高模型的鲁棒性。试验结果表明，YOLOX-RoC的AP比基准模型提升2.4百分点，针对小尺寸和被严重遮挡的小麦穗目标可以更准确地预测目标边界和角度，减少漏检和误检。本研究为小麦穗目标检测提供了一种准确和鲁棒的解决方案，为小麦估产和育种的智能化奠定了技术基础。
Abstract:
-

参考文献/References:

［1］Zhao C J，Zhang Y，Du J J，et al. Crop phenomics：current status and perspectives［J］. Frontiers in Plant Science，2019，10：714.
［2］Girshick R. Fast R-CNN［C］//2015 IEEE International Conference on Computer Vision. Santiago：IEEE，2015：1440-1448.
［3］Redmon J，Divvala S，Girshick R，et al. You only look once：unified，real-time object detection［C］//2016 IEEE Conference on Computer Vision and Pattern Recognition.Las Vegas：IEEE，2016：779-788.
［4］Liu W，Anguelov D，Erhan D，et al. SSD：single shot MultiBox detector［M］//Lecture Notes in Computer Science.Cham：Springer International Publishing，2016：21-37.
［5］Hasan M M，Chopin J P，Laga H，et al. Detection and analysis of wheat spikes using Convolutional Neural Networks［J］. Plant Methods，2018，14：100.
［6］He M X，Hao P，Xin Y Z.A robust method for wheatear detection using UAV in natural scenes［J］. IEEE Access，2020，8：189043-189053.
［7］刘航，刘涛，李世娟，等. 基于深度残差网络的麦穗回归计数方法［J］. 中国农业大学学报，2021，26(6)：170-179.
［8］黄子琦. 基于深度学习麦穗识别的小麦估产研究［D］. 泰安：山东农业大学，2022：3-4.
［9］Ge Z，Liu S，Wang F，et al. Yolox：Exceeding yolo series in 2021［EB/OL］. (2021-07-18)［2023-10-05］. https：//arxiv.org/abs/2107.08430.
［10］Redmon J，Farhadi A.YOLO v3：an incremental improvement［EB/OL］. (2018-04-08)［2023-10-05］. https：//arxiv.org/abs/1804.02767v1.
［11］Ioffe S，Szegedy C. Batch normalization：accelerating deep network training by reducing internal covariate shift［J］. 32nd International Conference on Machine Learning，2015，1：448-456.
［12］Xu B，Wang N Y，Chen T Q，et al. Empirical evaluation of rectified activations in convolutional network［EB/OL］. (2015-11-27)［2023-10-05］. https：//arxiv.org/abs/1505.00853v2.
［13］He K M，Zhang X Y，Ren S Q，et al. Spatial pyramid pooling in deep convolutional networks for visual recognition［J］. IEEE Transactions on Pattern Analysis and Machine Intelligence，2015，37(9)：1904-1916.
［14］Huang Z C，Wang J L，Fu X S，et al. DC-SPP-YOLO：dense connection and spatial pyramid pooling based YOLO for object detection［J］. Information Sciences，2020，522：241-258.
［15］Hu J，Shen L，Sun G.Squeeze-and-excitation networks［C］//2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition.Salt Lake：IEEE，2018：7132-7141.
［16］Woo S，Park J，Lee J Y，et al. CBAM：convolutional block attention module［M］//Lecture Notes in Computer Science.Cham：Springer International Publishing，2018：3-19.
［17］Hou Q B，Zhou D Q，Feng J S.Coordinate attention for efficient mobile network design［C］//2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition.Nashville：IEEE，2021：13708-13717.
［18］Yang X，Yang X J，Yang J R，et al. Learning high-precision bounding box for rotated object detection via kullback-leibler divergence［EB/OL］. (2021-06-03)［2023-10-05］. https：//arxiv.org/abs/2106.01883v5.
［19］Zakaria Y，Mokhtar S A，Baraka H，et al. Improving small and cluttered object detection by incorporating instance level denoising into single-shot alignment network for remote sensing imagery［J］. IEEE Access，2022，10：51176-51190.
［20］David E，Serouart M，Smith D，et al. Global Wheat Head Dataset 2021：more diversity to improve the benchmarking of wheat head localization methods［EB/OL］. (2021-06-03)［2023-10-05］. https：//arxiv.org/abs/2105.07660v2.
［21］Madec S，Jin X L，Lu H，et al. Ear density estimation from high resolution RGB imagery using deep learning technique［J］. Agricultural and Forest Meteorology，2019，264：225-234.
［22］刘颖，刘红燕，范九伦，等. 基于深度学习的小目标检测研究与应用综述［J］. 电子学报，2020，48(3)：590-601.
［23］Ghiasi G，Cui Y，Srinivas A，et al. Simple copy-paste is a strong data augmentation method for instance segmentation［C］//2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Nashville：IEEE，2021：2918-2928.

相似文献/References:

[1]罗巍,陈曙东,王福涛,等.基于深度学习的大型食草动物种群监测方法[J].江苏农业科学,2020,48(20):247.
　Luo Wei,et al.Monitoring method of large herbivore population based on deep learning[J].Jiangsu Agricultural Sciences,2020,48(20):247.
[2]陈恩会,褚姝频,王炜,等.基于RetinaNet模型的梨小食心虫智能识别计数方法[J].江苏农业科学,2021,49(24):205.
　Chen Enhui,et al.Intelligent recognition and counting method of Grapholitha molesta based on RetinaNet model[J].Jiangsu Agricultural Sciences,2021,49(20):205.
[3]陶雪阳,施振旦,郭彬彬,等.基于RFID与目标检测的种鹅个体产蛋信息监测方法[J].江苏农业科学,2023,51(5):200.
　Tao Xueyang,et al.Monitoring method of individual egg-laying information of breeding geese based on RFID and object detection[J].Jiangsu Agricultural Sciences,2023,51(20):200.
[4]严陈慧子,田芳明,谭峰,等.基于改进YOLOv4的水稻病害快速检测方法[J].江苏农业科学,2023,51(6):187.
　Yanchen Huizi,et al.Rapid detection method of rice diseases based on improved YOLOv4[J].Jiangsu Agricultural Sciences,2023,51(20):187.
[5]周绍发,肖小玲,刘忠意,等.改进的基于YOLOv5s苹果树叶病害检测[J].江苏农业科学,2023,51(13):212.
　Zhou Shaofa,et al.Improved apple leaf disease detection based on YOLOv5s[J].Jiangsu Agricultural Sciences,2023,51(20):212.
[6]姜国权,杨正元,霍占强,等.基于改进YOLOv5网络的疏果前苹果检测方法[J].江苏农业科学,2023,51(14):205.
　Jiang Guoquan,et al.Apple detection method before thinning fruit based on improved YOLOv5 model[J].Jiangsu Agricultural Sciences,2023,51(20):205.
[7]王圆圆,林建,王姗.基于YOLOv4-tiny模型的水稻早期病害识别方法[J].江苏农业科学,2023,51(16):147.
　Wang Yuanyuan,et al.An early rice disease recognition method based on YOLOv4-tiny model[J].Jiangsu Agricultural Sciences,2023,51(20):147.
[8]倪智涛,胡伟健,李宝山,等.一种基于图像分类与目标检测协同的番茄细粒度病害识别方法[J].江苏农业科学,2023,51(22):221.
　Ni Zhitao,et al.A novel method for tomato fine-grained disease recognition based on image classification and target detection[J].Jiangsu Agricultural Sciences,2023,51(20):221.
[9]施杰,林双双,罗建刚,等.基于YOLO v5s改进模型的玉米作物病虫害检测方法[J].江苏农业科学,2023,51(24):175.
　Shi Jie,et al.Study on a detection method for crop diseases and insect pests based on YOLO v5s improved model[J].Jiangsu Agricultural Sciences,2023,51(20):175.
[10]郑旭康,李志忠,秦俊豪.基于半监督学习的梨叶病害检测[J].江苏农业科学,2024,52(5):192.
　Zheng Xukang,et al.Study on pear leaf disease detection based on semi-supervised learning[J].Jiangsu Agricultural Sciences,2024,52(20):192.

备注/Memo

备注/Memo:
收稿日期：2023-11-06
基金项目：新疆维吾尔自治区重大科技专项(编号：2022A02011)。
作者简介：张世豪(1998—)，男，山东临沂人，硕士研究生，研究方向为深度学习与计算机视觉。E-mail：320203293@xjau.edu.cn。
通信作者：董峦，博士，副教授，研究方向为深度学习与计算机视觉。E-mail：dl@xjau.edu.cn。
更新日期/Last Update: 2024-10-20