基于YOLOv5的机耕船双目视觉障碍感知研究

doi:10.13733/j.jcam.issn.2095-5553.2024.07.039

摘要/Abstract

摘要： 为满足机耕船自动驾驶功能，设计一套YOLOv5融合SGBM算法的机器视觉障碍感知系统。首先，以人、机耕船和农具为对象拍摄和收集图片得到水田障碍数据集，将图像输入YOLOv5网络模型迭代训练得到最优权重，随后将最优权重用于测试，并且与YOLOv4和Faster R-CNN网络进行比较；将双目相机拍摄的左右图像输入YOLOv5模型中进行检测，将输出的目标障碍检测框信息经校正变换后用SGBM算法进行视差计算，完成对目标障碍的识别和深度估计。结果表明，YOLOv5的平均精度均值稳定在87.25%比YOLOv4高1.55%，比Faster R-CNN高4.04%，单幅图像检测时间为0.017 s比YOLOv4快0.081 s，比Faster R-CNN快0.182 s且模型大小仅为13.7 MB比YOLOv4小236.4 MB；在检测机耕船、人和农具时，YOLOv5网络模型的置信度分别为0.91、0.99、0.95。YOLOv5+SGBM的深度估计在2 m内，精度达到98.1%。基于YOLOv5和SGBM的水田深度估计，能满足带旋耕无人驾驶作业的机耕船实际需求。

关键词: 机耕船, 障碍感知, 机器视觉, YOLOv5, 深度估计

Abstract: In order to satisfy the automatic driving function of the boat tractor, this paper designed a set of YOLOv5 integrated SGBM algorithm machine vision obstacle perception system. Firstly, people, machinetiller and farm tools were taken as objects to shoot and collect images to get paddy field obstacle data set. The images were input into the YOLOv5 network model for iterative training to get the optimal weight. Later, the most weight was used for testing and compared with YOLOv4 and Faster R-CNN networks. The left and right images taken by the binocular camera were input into the YOLOv5 model for detection. After correcting and transforming the output information of the target obstacle detection box, the SGBM algorithm was used for parallax calculation to complete the target obstacle recognition and depth estimation. The results show that the average accuracy of YOLOv5 is stable at 87.25%, 1.55% higher than that of YOLOv4, 4.04% higher than that of Faster R-CNN, and the detection time of a single image is 0.017 s, 0.081 s faster than that of YOLOv4. It is 0.182 s faster than Faster R-CNN, and the model size is only 13.7 MB, 236.4 MB smaller than YOLOv4. The confidence of the YOLOv5 network model is 0.91, 0.99 and 0.95 respectively when detecting the boat tractor, man and farm tools. The depth estimation of YOLOv5+SGBM within 2 m, and the accuracy reaches 98.1%. The paddy field depth estimation based on YOLOv5 and SGBM can meet the actual requirements of unmanned boat tractor with rotary tillage.

Key words: boat tractor, obstacles perception, machine vision, YOLOv5, depth estimation

中图分类号:

陈泉淦, 陈新元, 曾镛, 程志文. 基于YOLOv5的机耕船双目视觉障碍感知研究[J]. 中国农机化学报, 2024, 45(7): 261-268.

Chen Quangan, , Chen Xinyuan, , Zeng Yong, , Cheng Zhiwen. Research on binocular visual impairment perception of a cultivator boat based on YOLOv5[J]. Journal of Chinese Agricultural Mechanization, 2024, 45(7): 261-268.

参考文献

［1］　夏俊芳, 张国忠, 许绮川, 等. 多熟制稻作区水田旋耕埋草机的结构与性能［J］. 华中农业大学学报, 2008, 27(2): 331-334.
Xia Junfang, Zhang Guozhong, Xu Qichuan, et al. Research on the mechanized technology of rotary tillage and stubblemulch for paddy field under multiple rice cropping system ［J］. Journal of Huazhong Agricultural University, 2008, 27(2): 331-334.
［2］　谭芳林. 机械法治理互花米草效果及其对滩涂土壤性状影响研究［J］. 湿地科学, 2008, 6(4): 526-530.
Tan Fanglin. Effect of the method combining cutting with machine boat on controlling spartina alterniflora and its impact on wetland soil characteristics ［J］. Wetland Science, 2008, 6(4): 256-530.
［3］　王峰. 浅析现代农机装备技术发展与农机维修工程［J］. 南方农机, 2022, 53(10): 165-167.
［4］　许绮川. 加快湖北省机耕船产业发展的建议［J］. 湖北农机化, 2014(4): 9-11.
［5］　姬长英, 沈子尧, 顾宝兴, 等. 基于点云图的农业导航中障碍物检测方法［J］. 农业工程学报, 2015, 31(7): 173-179.
Ji Changying, Shen Ziyao, Gu Baoxing, et al. Obstacle detection based on point clouds in application of agricultural navigation ［J］. Transactions of the Chinese Society of Agricultural Engineering, 2015, 31(7): 173-179.
［6］　薛金林, 董淑娴, 范博文. 基于信息融合的农业自主车辆障碍物检测方法［J］. 农业机械学报, 2018, 49(S1): 29-34.
Xue Jinlin, Dong Shuxian, Fan Bowen. Detection of obstacles based on information fusion for autonomous agricultural vehicles ［J］. Transactions of the Chinese Society for Agricultural Machinery, 2018, 49(S1): 29-34.
［7］　李乐, 张茂军, 熊志辉, 等. 基于内容理解的单幅静态街道景图像深度估计［J］. 机器人, 2011, 33(2): 174-180.
Li Le, Zhang Maojun, Xiong Zhihui, et al. Depth estimation from a single still image of street scene based on content understanding ［J］. Robot, 2011, 33(2): 174-180.
［8］　丁伟利, 李勇, 王文锋, 等. 基于轮廓特征理解的城市道路图像深度估计［J］. 光学学报, 2014, 34(7): 173-179.
Ding Weili, Li Yong, Wang Wenfeng, et al. Depth estimation of urban road image based on contour understanding ［J］. Acta Optica Sinica, 2014, 34(7): 173-179.
［9］　查志华, 周文静, 吴杰. 基于迁移学习Faster R-CNN模型田间红提葡萄果穗的识别［J］. 石河子大学学报(自然科学版), 2021, 39(1): 26-31.
Cha Zhihua, Zhou Wenjing, Wu Jie. Identification of red globe grape cluster in grapery with Faster R-CNN model based on transfer learning ［J］. Journal of Shihezi University (Natural Science), 2021, 39(1): 26-31.
［10］　黄华毅, 马晓航, 扈丽丽, 等. Fast R-CNN深度学习和无人机遥感相结合在松材线虫病监测中的初步应用研究［J］. 环境昆虫学报, 2021, 43(5): 1295-1303.
［11］　Shi Q, Li C, Guo B, et al. Manipulatorbased autonomous inspections at road checkpoints: Application of faster YOLO for detecting large objects ［J］. Defence Technology, 2022, 18(6): 937-951.
［12］　王淑青, 顿伟超, 黄剑锋, 等. 基于YOLOv5的瓷砖表面缺陷检测［J］. 包装工程, 2022, 43(9): 217-224.
Wang Shuqing, Dun Weichao, Huang Jianfeng, et al. Ceramic tile surface defect detection based on YOLOv5 ［J］. Packaging Engineering, 2022, 43(9): 217-224.
［13］　刘建政, 梁鸿, 崔学荣, 等. 融入特征融合与特征增强的SSD目标检测［J］. 计算机工程与应用, 2022, 58(11): 150-159.
Liu Jianzheng, Liang Hong, Cui Xuerong, et al. SSD visual target detector based on feature integration and feature enhancement ［J］. Computer Engineering and Applications, 2022, 58(11): 150-159.
［14］　齐继超, 何丽, 袁亮, 等. 基于单目相机与激光雷达融合的SLAM方法［J］. 电光与控制, 2022, 29(2): 99-102, 112.
Qi Jichao, He Li, Yuan Liang, et al. SLAM method based on fusion of monocular camera and Lidar camera and Lidar ［J］. Electronics Optics & Control, 2022, 29(2): 99-102, 112.
［15］　魏建胜, 潘树国, 田光兆, 等. 农业车辆双目视觉障碍物感知系统设计与试验［J］. 农业工程学报, 2021, 37(9): 55-63.
Wei Jiansheng, Pan Shuguo, Tian Guangzhao, et al. Design and experiments of the binocular visual obstacle perception system for agricultural vehicles ［J］. Transactions of the Chinese Society of Agricultural Engineering, 2021, 37(9): 55-63.
［16］　Yu Zihao, Liu Jin, Yang Haima, et al. Threedimensional surface reconstruction based on edge detection and reliability sorting algorithm ［J］. Laser & Optoelectronics Progress, 2020, 57(24).
［17］　Rezatofighi H, Tsoi N, Gwak J Y, et al. Generalized intersection over union: A metric and a loss for bounding box regression ［C］. 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2019.
［18］　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.
［19］　Zhang Z Y. A flexible new technique for camera calibration ［J］. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2000, 22(11): 1330-1334.

[1]	桑一男 , 徐增莱 , 汪琼 , 葛海涛 , 王殿广 . 一种轮式采花机器人设计——以黄蜀葵为例[J]. 中国农机化学报, 2024, 45(9): 202-208.
[2]	柳军, 孔杰, 皮杰, 周成钢. 果蔬分拣设备研究现状及发展趋势[J]. 中国农机化学报, 2024, 45(8): 120-125.
[3]	赵晓霞, 程曼, 袁洪波. 基于StrongSORT算法的羊只多目标跟踪方法[[J]. 中国农机化学报, 2024, 45(8): 180-188.
[4]	许贞辉, 李晓娟. 果园树干检测与导航线拟合算法研究[J]. 中国农机化学报, 2024, 45(8): 217-222.
[5]	冯晓, 张辉, 刘运超, 张微, 李小红, 马中杰, . 基于改进YOLOv5s的硬核期葡萄簇检测[J]. 中国农机化学报, 2024, 45(8): 240-245.
[6]	周思捷, 刘天奇, 陈天华. 基于改进YOLOv5算法的水稻病害识别研究[J]. 中国农机化学报, 2024, 45(8): 246-253.
[7]	王克涛, 陈世锋, 陈贵, 韦锦, 蒙丽雯, 陈泉成. 基于YOLOv5s的园林球形绿篱检测方法[J]. 中国农机化学报, 2024, 45(8): 262-269.
[8]	王雷, 钟康生, 郭小宝, 盛斌, 肖波, 荆磊. 基于机器视觉的陈皮热泵干燥含水率预测模型研究[J]. 中国农机化学报, 2024, 45(7): 97-103.
[9]	岳耀华, 张伟, 亓立强, . 基于改进YOLOv5的田间大豆花朵生长状态检测方法研究[J]. 中国农机化学报, 2024, 45(7): 188-193.
[10]	张延军, 赵建鑫, . 基于多路神经网络的黄花菜关节点检测算法研究[J]. 中国农机化学报, 2024, 45(7): 228-234.
[11]	吴利刚, 吕媛媛, 周倩, 陈乐, 张梁, 史建华. 基于改进YOLOv5的轻量级黄花成熟检测方法[J]. 中国农机化学报, 2024, 45(7): 235-242.
[12]	杨昊霖, 王其欢, 李华彪, 耿端阳, 武继达, 姚艳春. 基于改进YOLOv5的田间复杂环境障碍物检测[J]. 中国农机化学报, 2024, 45(6): 216-222.
[13]	黄尧, 何敬, 付饶, 刘刚, , 林远杨. YOLOv5s-CBAM算法在福寿螺虫卵识别中的应用分析[J]. 中国农机化学报, 2024, 45(6): 223-228.
[14]	周康乔, 刘向阳, 郑特驹. 改进YOLOv5的小目标多类别农田害虫检测算法研究[J]. 中国农机化学报, 2024, 45(6): 235-241.
[15]	严蓓蓓, 纪元浩, 曲凤凤, 许金普. 基于改进YOLOv5s的茶叶嫩芽检测[J]. 中国农机化学报, 2024, 45(4): 168-174.