基于深度学习的林下落果识别方法与试验

doi:10.13733/j.jcam.issn.2095-5553.2021.09.28

摘要/Abstract

摘要： 监测与识别林下落果的数量和分布信息,是实现落果自动收获和果园智能化管理的重要基础。针对目前落果识别智能化程度较低等问题,提出一种基于深度学习的林下落果识别方法。首先,以不同类型、品种落果图像为基础,通过数据预处理、增强等方法建立林下落果图像数据集。其次,利用YOLO-v3深度卷积神经网络优势特性,建立落果智能识别方法。最后,以柑橘、梨、苹果三种典型落果,对基于深度学习的林下落果识别方法进行测试与验证,分析相关试验结果。试验结果表明:所提出的基于YOLO-v3落果识别方法,在不同条件均能准确识别落果,三种典型落果识别精度大于89%;相对于SSD,RCNN和CenterNet三种网络模型,YOLO-v3的准确率分别提高7%,2%和3.5%;在腐烂落果识别层面,YOLO-v3、SSD、RCNN和CenterNet的识别准确率分别为86%,59%,64%和43%;YOLO-v3的识别准确率高于其他深度学习模型。所提出的方法可以精确的识别林下落果,为后期的落果精准管理提供必要的技术支撑。

关键词: 落果识别, 目标检测, 图像处理, 深度神经网络

Abstract: In order to monitor the quantity and distribution and provide the necessary support for automatic harvesting of the fallen fruit and intelligent orchard management, this paper proposes a deep learningbased fruit fall identification method to solve the problem of low automatic monitoring degree of fruit fall at present. Firstly, the fallen fruit image dataset was established with consideration of different images by using data collection and preprocessing. Secondly, YOLOv3 (You Only Look Once) deep convolutional neural networks were used to identify the fallen fruits. Finally, an experiment was conducted toevaluate the performance of the proposed method. The results showed that the proposed algorithm could accurately identify fallen fruit under different types and different occlusion conditions. YOLOv3 got the most value (more than 89%) of the accuracy of indentation. Compared with SSD, RCNN, and CenterNet, the accuracy of YOLOv3 increasedaccuracy by 7%, 2%, and 3.5%, respectively. The recognition accuracy of rotting fallen fruits by using YOLOv3, SSD, RCNN, and CenterNetreached 86%, 59%, 64%, and 43%, respectively. The recognition accuracy of YOLOv3 was higher than that of other deep learning models. The proposed method can accurately identify the number of the fallen fruit and provide necessary technical support for accurate management of fallen fruit.

Key words: fruit drop recognition, target detection, image processing, deep neural network

中图分类号:

李小敏, 张日红, 陈天赐, 侯炳法, 张权, 李雄, . 基于深度学习的林下落果识别方法与试验[J]. 中国农机化学报, 2021, 42(9): 202-208.

Li Xiaomin, Zhang Rihong, Chen Tianci, Hou Bingfa, Zhang Quan, Li Xiong.. Identification method and experiment of fallen fruit based on deep learning[J]. Journal of Chinese Agricultural Mechanization, 2021, 42(9): 202-208.

参考文献

［１］　
李秀根. 近年我国梨产业发展中遇到的陷阱与建议［J］. 果农之友, 2020(12): 1-3.

［2］　
陈哲, 梁瀚元. 我国西瓜产业发展与贸易趋势研究［J］. 经营与管理, 2019(9): 114-122.

［3］　
张放. 2018年我国主要水果生产统计简析［J］. 中国果业信息, 2020, 37(7): 32-43.

［4］　
张放. 2017年我国主要水果生产统计分析［J］. 中国果业信息, 2019, 36(10): 22-31.

［5］　
许培磊, 韩先焱, 李昌禹, 等. 果树落花落果机制研究进展［J］. 特产研究, 2018, 40(4): 82-91.

Xu Peilei, Han Xianyan, Li Changyu, et al. Overview of flower and fruit drop mechanism of fruit trees ［J］. Special Wild Economic Animal and Plant Research, 2018, 40(4): 82-91.

［6］　
太一梅, 朱斌, 何绍华, 等. 杨梅果蝇综合防治技术研究［J］. 植物保护, 2015, 41(1): 193-195.

Tai Yimei, Zhu Bin, He Shaohua, et al. Integrated control techniques of fruit fly (Drosophila melanogaster Meigen) in waxberry ［J］. Plant Protection, 2015, 41(1): 193-195.

［7］　
Khamis Y, Roberto S R. Premature apple fruit drop: associated fungal species and attempted management solutions ［J］. Horticulturae, 2020, 6(2): 31.

［8］　
Fabrícìo E. Lanza, Weber Marti, Geraldo J. Silva, et al. Characteristics of citrus canker lesions associated with premature drop of sweet orange fruit ［J］. Phytopathology. 2019, 109(1): 44-51.

［9］　
Kumar T B, Kumar G P, Kumar R S, et al. Effect of nutrient management through bioorganic manures on fruit setting, fruit drop and fruit retention of acid lime (Citrus Aurantifolia Swingle) ［J］. Plant Archives, 2020, 20(1): 1570-1572.

［10］　
张利军, 罗自威, 王玉雯, 等. 琯溪蜜柚落花落果特性及养分损失定量化研究［J］. 果树学报, 2020, 38(4): 520-529.

Zhang Lijun, Luo Ziwei, Wang Yuwen, et al. A study on the characteristic of dropped flower and fruit and its nutrient loss quantification in Guanximiyou pomelo ［J］. Journal of Fruit Science, 2020, 38(4): 520-529.

［11］　
杨小慧, 石光波, 拜晓彬, 等. 文冠果落果黄酮成分分析及抑菌性评价［J］. 食品科学, 2018, 39(10): 53-58.

Yang Xiaohui, Shi Guangbo, Bai Xiaobin, et al. Flavonoid content and antibacterial activity of Xanthoceras sorbifolia Bunge fruit drop ［J］. Food Science, 2018, 39(10): 53-58.

［12］　
Guijarro M, Pajares G, Riomoros I, et al. Automatic segmentation of relevant textures in agricultural images ［J］. Computers and Electronics in Agriculture, 2011, 75(1): 75-83.

［13］　
Konstantinos L, Patrizia B, Dimitrios M, et al. Machine learning in agriculture: A review ［J］. Sensors, 2018, 18(8): 2674.

［14］　
Tang J L, Wang D, Zhang Z G, et al. Weed identification based on K-means feature learning combined with convolutional neural network ［J］. Computers & Electronics in Agriculture, 2017, 135: 63-70.

［15］　
Lebrini Y, Boudhar A, Hadria R, et al. Identifying agricultural systems using SVM classification approach based on phenological metrics in a semiarid region of Morocco ［J］. Earth Systems and Environment, 2019, 3(2): 277-288.

［16］　
Kamilaris A, PrenafetaBoldu F X. Deep learning in agriculture: A survey ［J］. Computers and Electronics in Agriculture, 2018, 147: 70-90.

［17］　
Zhu N, Xu L, Liu Z, et al. Deep learning for smart agriculture: Concepts, tools, applications, and opportunities ［J］. International Journal of Agricultural and Biological Engineering, 2018, 11(4): 21-28.

［18］　
Kim S, Peter C, Henrik K, et al. Using deep learning to challenge safety standard for highly autonomous machines in agriculture ［J］. Journal of Imaging, 2016, 2(1): 6.

［19］　
Nguyen D T, Nguyen T N, Kim H, et al. A highthroughput and powerefficient FPGA implementation of YOLO CNN for object detection ［J］. IEEE Transactions on Very Large Scale Integration (VLSI) Systems, 2019, 27(8): 1861-1873.

［20］　
Hendry, Chen R C. Automatic license plate recognition via SlidingWindow DarknetYolo deep learning ［J］. Image and Vision Computing, 2019, 87: 47-56.

［21］　
Du J. Understanding of object detection based on CNN family and YOLO ［C］. Journal of Physics: Conference Series. IOP Publishing, 2018, 1004(1): 012029.

［22］　
Tian Y, G Yang, Wang Z, et al. Apple detection during different growth stages in orchards using the improved YOLOv3 model ［J］. Computers and electronics in agriculture, 2019, 157: 417-426.

［23］　
Lawal M O. Tomato detection based on modified YOLOv3 framework ［J］. Scientific Reports, 2021, 11(1): 1-11.

［24］　
熊俊涛, 郑镇辉, 梁嘉恩, 等. 基于改进YOLO v3网络的夜间环境柑橘识别方法［J］. 农业机械学报, 2020, 51(4): 199-206.

Xiong Juntao, Zheng Zhenhui, Liang Jiaen, et al. Citrus detection method in night environment based on improved YOLO v3 Network ［J］. Transactions of the Chinese Society for Agricultural Machinery, 2020, 51(4): 199-206.

［25］　
荆伟斌, 胡海棠, 程成, 等. 基于深度学习的地面苹果识别与计数［J］. 江苏农业科学, 2020, 48(5): 210-219.

[1]	梁喜凤, 花瑞, 杨铭涛. 番茄枝叶修剪机械手视觉伺服控制系统研究[J]. 中国农机化学报, 2022, 43(7): 8-13.
[2]	宋中山, 汪进, 郑禄, 帖军, 朱祖桐, . 基于二值化的Faster R-CNN柑橘病虫害识别研究[J]. 中国农机化学报, 2022, 43(6): 150-158.
[3]	史方青, 王虎林, 黄华. 基于卷积神经网络的马铃薯芽眼检测识别研究[J]. 中国农机化学报, 2022, 43(6): 159-165.
[4]	阮承治, 陈茜, 陈旭, 赵升云, 孙月平, 赵德安. 颜色空间直方图相交法的竹条色差分类检测系统研究[J]. 中国农机化学报, 2022, 43(5): 85-92.
[5]	李金阳, 张伟, 康烨, 许秀英, 亓立强, 石文强, . 基于无人机遥感技术的大豆苗数估算研究[J]. 中国农机化学报, 2022, 43(4): 83-89.
[6]	何颖, 陈丁号, 彭琳. 基于改进YOLOv5模型的经济林木虫害目标检测算法研究[J]. 中国农机化学报, 2022, 43(4): 106-115.
[7]	郑如新, 孙青云, 肖国栋. 基于机器视觉的金银花图像识别处理算法研究[J]. 中国农机化学报, 2022, 43(4): 153-159.
[8]	赵守耀, 陆辉山, 王福杰, 李沛, 王宁. 基于轮廓特征的单只蛋鸡行为识别方法[J]. 中国农机化学报, 2022, 43(2): 143-147.
[9]	王菁, 范晓飞, 赵智慧, 张君, 孙磊, 索雪松. 基于YOLO算法的不同品种枣自然环境下成熟度识别[J]. 中国农机化学报, 2022, 43(11): 165-171.
[10]	王雪晴;郑福来;赵换丽;. 基于几何和色度学特征识别的花生视觉选种方法[J]. 中国农机化学报, 2021, 42(5): 94-99.
[11]	项新建;周跃琪;姚佳娜;宋晓敏;郑永平;. 野外农用视频监控运动目标检测算法研究及系统开发[J]. 中国农机化学报, 2021, 42(5): 166-174.
[12]	郭祥雨;薛新宇;. 基于机器视觉的水稻制种田导航线提取方法[J]. 中国农机化学报, 2021, 42(5): 197-201.
[13]	成伟;张文爱;冯青春;张万豪;. 基于改进YOLOv3的温室番茄果实识别估产方法[J]. 中国农机化学报, 2021, 42(4): 176-182.
[14]	闫建伟;张乐伟;赵源;张富贵;. 改进RetinaNet的刺梨果实图像识别[J]. 中国农机化学报, 2021, 42(3): 78-83.
[15]	单慧勇;张程皓;李晨阳;赵辉;田云臣;. 鱼加工生产线头尾定向调理上料系统设计与试验[J]. 中国农机化学报, 2021, 42(3): 91-97.