基于实例分割的叶菜垄间导航线提取算法研究

doi:10.3969/j.issn.1004-1524.20240167

浙江农业学报 ›› 2025, Vol. 37 ›› Issue (3): 701-711.DOI: 10.3969/j.issn.1004-1524.20240167

基于实例分割的叶菜垄间导航线提取算法研究

郑航¹^,²(), 冯昊栋³, 薛向磊¹^,², 叶云翔¹^,², 于健麟¹, 俞国红¹^,²^,^*()

1.浙江省农业科学院农业装备研究所,浙江杭州 310021
2.农业农村部东南丘陵山地农业装备重点实验室(部省共建),浙江杭州 310021
3.浙江理工大学机械工程学院,浙江杭州 310018

收稿日期:2024-02-26 出版日期:2025-03-25 发布日期:2025-04-02
作者简介:郑航(1993—),男,浙江龙游人,博士,助理研究员,研究方向为智能农机装备设计与制造。E-mail:Zhrory@126.com
通讯作者: * 俞国红,E-mail:Yuguohong@163.com
基金资助:
浙江省“领雁”科技计划项目(2023C02053);浙江省农业科学院成果推广项目(2023R30CB001)

Study on navigation line extraction algorithm for leaf vegetable ridges based on instance segmentations

ZHENG Hang¹^,²(), FENG Haodong³, XUE Xianglei¹^,², YE Yunxiang¹^,², YU Jianlin¹, YU Guohong¹^,²^,^*()

1. Institute of Agricultural Equipment, Zhejiang Academy of Agricultural Sciences, Hangzhou 310021, China
2. Key Laboratory of Agricultural Equipment for Hilly and Mountainous Areas in Southeastern China (Co-Construction by Ministry and Province), Ministry of Agriculture and Rural Affairs, Hangzhou 310021, China
3. School of Mechanical Engineering, Zhejiang Sci-Tech University, Hangzhou 310018, China

Received:2024-02-26 Online:2025-03-25 Published:2025-04-02

摘要/Abstract

摘要：

为提高设施垄作模式下叶菜的田间管理装备移动导航精度,文章提出了一种基于深度学习的叶菜垄间导航线提取算法。首先在叶菜垄间导航路径数据集上进行模型训练,采用改进的YOLOv5s-Seg 卷积神经网络提取温室作业道路图像的特征点,并通过最小二乘法拟合生成导航线。在试验中,对模型的内部结构进行了一系列的改进,提高了算法的精度和运算速度;通过采集的800张垄作模式下叶菜类作物生长图片,并通过数据增强将数据集总数扩充为原来的4倍,进而在320张图片组成的独立测试集上进行测试,总体分割精度均值为99.50%,在发芽期、幼苗期和产品器官形成期,图像的准确率分别为94.33%、97.77%和96.23%,拟合导航线与人工观测导航线的平均偏差为5.60 cm。结果表明,该导航算法能满足设施内垄作模式叶菜智能化管理移动装备的导航需求。

关键词: 设施温室, 垄作模式, 实例分割, 深度学习, 导航线

Abstract:

To improve the mobile navigation accuracy of field management equipment for leafy vegetables under facility ridge cultivation mode, a deep learning based algorithm for extracting navigation lines between leafy vegetable ridges was proposed. Firstly, the model was trained on the navigation path dataset between leafy vegetable ridges. An improved YOLOv5s-Seg convolutional neural network was used to extract feature points from greenhouse operation road images, and the navigation line was generated through least squares fitting. In the experiment, a series of improvements were made to the internal structure of the model, which improved the accuracy and computational speed of the algorithm. By collecting 800 images of the growth of leafy vegetables under ridge planting mode, and expanding the total number of datasets to four times of the original through data augmentation, the overall segmentation average precision was 99.50% on an independent test set composed of 320 images. The precision of images were 94.33%, 97.77% and 96.23% respectively at germination stage, seedling stage, and formation stage. The average deviation between the fitted navigation line and the manually observed navigation line was 5.60 cm. The results showed that the navigation algorithm could meet the navigation requirements of intelligent management of mobile equipment for leafy vegetables in the ridge planting mode within the facility.

Key words: facility greenhouse, ridge planting mode, instance segmentation, deep learning, navigation line

中图分类号:

S225.92

郑航, 冯昊栋, 薛向磊, 叶云翔, 于健麟, 俞国红. 基于实例分割的叶菜垄间导航线提取算法研究[J]. 浙江农业学报, 2025, 37(3): 701-711.

ZHENG Hang, FENG Haodong, XUE Xianglei, YE Yunxiang, YU Jianlin, YU Guohong. Study on navigation line extraction algorithm for leaf vegetable ridges based on instance segmentations[J]. Acta Agriculturae Zhejiangensis, 2025, 37(3): 701-711.

图/表 16

图1 图像采集

Fig.1 Image acquisition

图2 叶菜不同生长期数据集

Fig.2 Dataset of leafy vegetables at different growth stages

图3 部分数据集样本

Fig.3 Partial dataset samples

图4 改进 YOLOv5-seg 模型

Fig.4 Improved YOLO5s-seg model

图5 坐标注意力机制流程图

Fig.5 Coordinate Attention mechanism

图6 PConv卷积结构

Fig.6 PConv convolution structure

图7 C3-Faster模块

Fig.7 C3-Faster convolution structure

图8 GhostConv卷积与普通卷积的比较

Fig.8 Comparison of GhostConv convolution and general convolution

图9 损伤函数曲线

Fig.9 Loss function curve

图10 验证集性能指标曲线

Fig.10 Verification set performance indicator curve

表1 改进后YOLOv5s-seg的训练结果

Table 1 Segmentation results of YOLOv5s-seg under different growth stages were improved %

生长时期 Growth stage	准确率 Precision	召回率 Recall	精度均值 Average precision
发芽期	94.33	95.92	97.25
Germination stage
幼苗期	97.77	96.52	98.49
Seedling stage
成形期	96.23	97.73	95.56
Formation stage

图11 YOLOv5-seg分割结果

Fig.11 YOLOv5-seg segmentation results

表2 消融试验结果

Table 2 Ablation test results

模型 Model	CA	Ghostconv	Pconv	Mask/%			Parameters	Flops
模型 Model	CA	Ghostconv	Pconv	准确率 Precision	召回率 Recall	精度均值 Average precision	Parameters	Flops
Yolov5s-seg				96.4	99.2	99.1	7 408 214	25.9
Yolov5s-seg+CA	√			98.5	96.9	99.4	7 421 550	25.9
Yolov5s-seg+Ghostconv		√		97.4	98.6	99.3	6 195 894	23.5
Yolov5s-seg+C3_Faster			√	98.1	97.4	99.4	6 177 686	22.8
本文算法Ours	√	√	√	99.6	99.7	99.5	4 978 702	20.4

表3 不同模型分割试验结果对比

Table 3 Comparison of the results of different model segmentation experiments

模型 Model	精度均值 Average precision/%	推理时间 Reasoning time/ms
Faster R-CNN	98.7	185.0
YOLOv7s-Seg	98.6	30.0
YOLOv8s-Seg	99.4	8.1
改进YOLOv5s-Seg	99.5	7.2
Improved YOLOv5s-Seg

图12 导航线生成流程

Fig.12 The navigation line generation process

图13 导航偏差分析

Fig.13 Navigation line deviation analysis

参考文献 25

[1]	丁亚会, 张云鹤, 孙宁, 等. 我国设施农业发展的国际经验与启示[J]. 江苏农业科学, 2023, 51(16): 1-8.
	DING Y H, ZHANG Y H, SUN N, et al. International experience and enlightenment on development of China’s facility agriculture[J]. Jiangsu Agricultural Sciences, 2023, 51(16): 1-8. (in Chinese with English abstract)
[2]	王江, 李浩, 马志伟, 等. 国内外设施蔬菜机械化发展现状分析及对策[J]. 中国农机化学报, 2023, 44(1): 124-130.
	WANG J, LI H, MA Z W, et al. Analysis and countermeasures of mechanization development of domestic and foreign facility vegetables[J]. Journal of Chinese Agricultural Mechanization, 2023, 44(1): 124-130. (in Chinese with English abstract)
[3]	胡静涛, 高雷, 白晓平, 等. 农业机械自动导航技术研究进展[J]. 农业工程学报, 2015, 31(10): 1-10.
	HU J T, GAO L, BAI X P, et al. Review of research on automatic guidance of agricultural vehicles[J]. Transactions of the Chinese Society of Agricultural Engineering, 2015, 31(10): 1-10. (in Chinese with English abstract)
[4]	刘成良, 林洪振, 李彦明, 等. 农业装备智能控制技术研究现状与发展趋势分析[J]. 农业机械学报, 2020, 51(1): 1-18.
	LIU C L, LIN H Z, LI Y M, et al. Analysis on status and development trend of intelligent control technology for agricultural equipment[J]. Transactions of the Chinese Society for Agricultural Machinery, 2020, 51(1): 1-18. (in Chinese with English abstract)
[5]	张漫, 季宇寒, 李世超, 等. 农业机械导航技术研究进展[J]. 农业机械学报, 2020, 51(4): 1-18.
	ZHANG M, JI Y H, LI S C, et al. Research progress of agricultural machinery navigation technology[J]. Transactions of the Chinese Society for Agricultural Machinery, 2020, 51(4): 1-18. (in Chinese with English abstract)
[6]	饶秀勤, 朱逸航, 张延宁, 等. 基于语义分割的作物垄间导航路径识别[J]. 农业工程学报, 2021, 37(20): 179-186.
	RAO X Q, ZHU Y H, ZHANG Y N, et al. Navigation path recognition between crop ridges based on semantic segmentation[J]. Transactions of the Chinese Society of Agricultural Engineering, 2021, 37(20): 179-186. (in Chinese with English abstract)
[7]	张雄楚, 陈兵旗, 李景彬, 等. 红枣收获机视觉导航路径检测[J]. 农业工程学报, 2020, 36(13): 133-140.
	ZHANG X C, CHEN B Q, LI J B, et al. Path detection of visual navigation for jujube harvesters[J]. Transactions of the Chinese Society of Agricultural Engineering, 2020, 36(13): 133-140. (in Chinese with English abstract)
[8]	宋宇, 刘永博, 刘路, 等. 基于机器视觉的玉米根茎导航基准线提取方法[J]. 农业机械学报, 2017, 48(2): 38-44.
	SONG Y, LIU Y B, LIU L, et al. Extraction method of navigation baseline of corn roots based on machine vision[J]. Transactions of the Chinese Society for Agricultural Machinery, 2017, 48(2): 38-44. (in Chinese with English abstract)
[9]	王宝聚, 兰玉彬, 陈蒙蒙, 等. 机器学习在无人农场中的应用现状与展望[J]. 中国农机化学报, 2021, 42(10): 186-192.
	WANG B J, LAN Y B, CHEN M M, et al. Application status and prospect of machine learning in unmanned farm[J]. Journal of Chinese Agricultural Mechanization, 2021, 42(10): 186-192. (in Chinese with English abstract)
[10]	兰玉彬, 闫瑜, 王宝聚, 等. 智能施药机器人关键技术研究现状及发展趋势[J]. 农业工程学报, 2022, 38(20): 30-40.
	LAN Y B, YAN Y, WANG B J, et al. Current status and future development of the key technologies for intelligent pesticide spraying robots[J]. Transactions of the Chinese Society of Agricultural Engineering, 2022, 38(20): 30-40. (in Chinese with English abstract)
[11]	高震宇, 王安, 刘勇, 等. 基于卷积神经网络的鲜茶叶智能分选系统研究[J]. 农业机械学报, 2017, 48(7): 53-58.
	GAO Z Y, WANG A, LIU Y, et al. Intelligent fresh-tea-leaves sorting system research based on convolution neural network[J]. Transactions of the Chinese Society for Agricultural Machinery, 2017, 48(7): 53-58. (in Chinese with English abstract)
[12]	王毅, 刘波, 熊龙烨, 等. 基于深度学习的果园道路导航线生成算法研究[J]. 湖南农业大学学报(自然科学版), 2019, 45(6): 674-678.
	WANG Y, LIU B, XIONG L Y, et al. Research on generating algorithm of orchard road navigation line based on deep learning[J]. Journal of Hunan Agricultural University(Natural Sciences), 2019, 45(6): 674-678. (in Chinese with English abstract)
[13]	韩振浩, 李佳, 苑严伟, 等. 基于U-Net网络的果园视觉导航路径识别方法[J]. 农业机械学报, 2021, 52(1): 30-39.
	HAN Z H, LI J, YUAN Y W, et al. Path recognition of orchard visual navigation based on U-Net[J]. Transactions of the Chinese Society for Agricultural Machinery, 2021, 52(1): 30-39. (in Chinese with English abstract)
[14]	孟庆宽, 张漫, 仇瑞承, 等. 基于改进遗传算法的农机具视觉导航线检测[J]. 农业机械学报, 2014, 45(10): 39-46.
	MENG Q K, ZHANG M, QIU R C, et al. Navigation line detection for farm machinery based on improved genetic algorithm[J]. Transactions of the Chinese Society for Agricultural Machinery, 2014, 45(10): 39-46. (in Chinese with English abstract)
[15]	吕艳鹏, 潘玉田. 基于BP神经网络的非结构化道路识别[J]. 机械工程与自动化, 2014(3): 178-180.
	LYU Y P, PAN Y T. Unstructured road detection based on BP neural network[J]. Mechanical Engineering & Automation, 2014(3): 178-180. (in Chinese with English abstract)
[16]	LIU F C, YANG Y, ZENG Y M, et al. Bending diagnosis of rice seedling lines and guidance line extraction of automatic weeding equipment in paddy field[J]. Mechanical Systems and Signal Processing, 2020, 142: 106791.
[17]	WANG S S, YU S S, ZHANG W Y, et al. The seedling line extraction of automatic weeding machinery in paddy field[J]. Computers and Electronics in Agriculture, 2023, 205: 107648.
[18]	YANG Y, ZHOU Y, YUE X, et al. Real-time detection of crop rows in maize fields based on autonomous extraction of ROI[J]. Expert Systems with Applications, 2023, 213: 118826.
[19]	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 (CVPR), Las Vegas, NV, USA. IEEE, 2016: 779-788.
[20]	MALTA A, MENDES M, FARINHA T. Augmented reality maintenance assistant using YOLOv5[J]. Applied Sciences, 2021, 11(11): 4758.
[21]	白卫卫, 赵雪妮, 罗斌, 等. 基于YOLOv5的小麦种子发芽检测方法研究[J]. 浙江农业学报, 2023, 35(2): 445-454.
	BAI W W, ZHAO X N, LUO B, et al. Study of YOLOv5-based germination detection method for wheat seeds[J]. Acta Agriculturae Zhejiangensis, 2023, 35(2): 445-454. (in Chinese with English abstract)
[22]	张河山, 范梦伟, 谭鑫, 等. 基于改进YOLOX的无人机航拍图像密集小目标车辆检测[J/OL]. 吉林大学学报(工学版),(2023-12-28)[2024-02-25]. http://kns-cnki-net-s.webvpn.zjda.net:8118/kcms2/article/abstract?v=tOz5m-jLbAVDO6YEQjcoGfIYPw7aUrQSxfdX_VGHBX6w0-bRG_MoC3MfRVHvmvRHZrLBLnybr3Z_a7HrCnBHlFDZsK-doI3Cs-2otyLNWOX9UF1xJ4o3pjniF NrOqAbEthiyODdX86-vfYNEq9K-p4D_QjNVed47O83KuTr7umGDfdaHd46BiiXAD-iwYbTks&uniplatform=NZKPT&language=CHS.
	ZHANG H S, FAN M W, TAN X, et al. Dense small object vehicle detection in UAV aerial images using improved YOLOX[J/OL]. Journal of Jilin University(Engineering and Technology Edition), (2023-12-28)[2024-02-25]. http://kns-cnki-net-s.webvpn.zjda.net:8118/kcms2/article/abstract?v=tOz5m-jLbAVDO6YEQjcoGfIYPw7aUrQSxfdX_VGHBX6w0-bRG_MoC3MfRVHvmvRHZrLBLnybr3Z_a7HrCnBHlFDZsKdoI3Cs-2otyLNWOX9UF1xJ4o3pjniFNrOqAbEthiyODdX86vfYNEq9K-p4D_QjNVed47O83KuTr7um-GDfdaHd46BiiXADiwYbTks&uniplatform=NZKPT&language=CHS. (in Chinese with English abstract)
[23]	ZHANG F, CAO W H, WANG S Q, et al. Improved YOLOv4 recognition algorithm for pitaya based on coordinate attention and combinational convolution[J]. Frontiers in Plant Science, 2022, 13: 1030021.
[24]	张继龙, 赵军, 李金龙. 改进YOLOv7的X光图像危险品检测算法[J]. 计算机工程与应用, 2024, 60(10):266-275.
	ZHANG J L, ZHAO J, LI J L. Improved dangerous goods detection in X-ray images of YOLOv7[J]. Computer Engineering and Applications, 2024, 60(10):266-275. (in Chinese with English abstract)
[25]	晏朋, 李遇鑫, 李治林, 等. 基于改进YOLOv7的高铁异物入侵检测算法[J]. 无线电工程, 2024, 54(5):1099-1109.
	YAN P, LL Y X, LL Z L, et al. High-speed rail foreign object intrusion detection algorithm based on improveo YOLOv7[J]. Radio Engineering, 2024, 54(5):1099-1109. (in Chinese with English abstract)

基于实例分割的叶菜垄间导航线提取算法研究

Study on navigation line extraction algorithm for leaf vegetable ridges based on instance segmentations

RichHTML

PDF (PC)

可视化

摘要/Abstract

引用本文

使用本文

图/表 16

参考文献 25

相关文章 10

编辑推荐

Metrics

本文评价

[1]	程嘉瑜, 陈妙金, 李彤, 孙奇男, 张小斌, 赵懿滢, 朱怡航, 顾清. 基于改进Faster-RCNN网络的无人机遥感影像桃树检测[J]. 浙江农业学报, 2024, 36(8): 1909-1919.
[2]	潘攀, 张建华, 郑晓明, 周国民, 胡林, 冯全, 柴秀娟. 深度学习在作物及其近缘种抗病性智能鉴定上的研究进展[J]. 浙江农业学报, 2023, 35(8): 1993-2012.
[3]	白卫卫, 赵雪妮, 罗斌, 赵薇, 黄硕, 张晗. 基于YOLOv5的小麦种子发芽检测方法研究[J]. 浙江农业学报, 2023, 35(2): 445-454.
[4]	朱世松, 马婉丽, 赵理山, 郑艳梅, 郑先波, 芦碧波. 基于改进的LinkNet的苹果叶片图像分割算法[J]. 浙江农业学报, 2023, 35(1): 202-214.
[5]	陈道怀, 汪杭军. 基于改进YOLOv4的林业害虫检测[J]. 浙江农业学报, 2022, 34(6): 1306-1315.
[6]	李超, 李锋, 黄炜嘉. 基于并联卷积神经网络的水果品种识别[J]. 浙江农业学报, 2022, 34(11): 2533-2541.
[7]	张晴晴, 刘连忠, 宁井铭, 吴国栋, 江朝晖, 李孟杰, 李栋梁. 基于YOLOV3优化模型的复杂场景下茶树嫩芽识别[J]. 浙江农业学报, 2021, 33(9): 1740-1747.
[8]	鲍烈, 王曼韬, 刘江川, 文波, 明月. 基于卷积神经网络的小麦产量预估方法[J]. 浙江农业学报, 2020, 32(12): 2244-2252.
[9]	王彦翔, 张艳, 杨成娅, 孟庆龙, 尚静. 基于深度学习的农作物病害图像识别技术进展[J]. 浙江农业学报, 2019, 31(4): 669-676.
[10]	杨国亮, 许楠, 康乐乐, 龚曼, 洪志阳. 基于参数指数非线性残差神经网络的脐橙病变叶片识别[J]. 浙江农业学报, 2018, 30(6): 1073-1081.