Computers and Electronics in Agriculture最新文献_第5页

Research on high-speed and clean production with a high-speed centrifugal maize precision seed metering device featuring variable hole insert numbers 利用具有可变插入孔数的高速离心玉米精密种子计量装置进行高速清洁生产的研究

IF 7.7 1区农林科学 Q1 AGRICULTURE, MULTIDISCIPLINARY

Computers and Electronics in Agriculture

Pub Date : 2024-11-12 DOI: 10.1016/j.compag.2024.109620

Chuan Li , Dongxing Zhang , Li Yang , Tao Cui , Xiantao He , Zhimin Li , Jiaqi Dong , Shulun Xing , Yeyuan Jiang , Jiyuan Liang

Traditional pneumatic seed metering devices rely on air pressure for seed filling and carrying, resulting in high energy consumption and limited seeding speed. While centrifugal seed metering devices can achieve high-speed seeding, they have a narrow optimal seeding speed range, making low-speed seeding difficult. In this study, the number of hole inserts in a high-speed centrifugal precision seed metering device for maize was set to 2, 4, 6, and 8, enabling precision seeding at higher speeds and across a broader speed range. Different agitator wheel structures were designed based on the number of hole inserts. The motion characteristics of the gas and seeds were analyzed using a combination of Discrete Element Method and Computational Fluid Dynamics to determine the optimal agitator wheel structure. Bench test results indicated that the optimal seeding speed ranges for 2, 4, 6, and 8 hole inserts were 6–9 km/h, 12–18 km/h, 18–27 km/h, and 24–36 km/h, respectively. With 8 hole inserts, the maximum seeding speed reached 36 km/h, achieving a miss rate of 2.75 %, a repeat rate of 3.76 %, and a qualification rate of 93.49 %. The energy consumption of the high-speed centrifugal maize precision seed metering device during seeding was less than 411.71 kJ/ha, which is less than 9 % of the energy consumed per hectare by pneumatic seed metering devices. Additionally, the higher the seeding speed, the lower the energy consumption per hectare. At a seeding speed of 36 km/h, the energy consumption was 90.08 kJ/ha. Compared to pneumatic seed metering devices, the high-speed centrifugal maize precision seed metering device offers higher seeding speeds and lower energy consumption, enabling high-speed and clean production.

传统的气动式种子计量装置依靠气压进行充种和运种，能耗高，播种速度有限。离心式种子计量装置虽然可以实现高速播种，但其最佳播种速度范围较窄，难以实现低速播种。在这项研究中，玉米高速离心式精密种子计量装置中的穴盘数被设置为 2、4、6 和 8，从而能够在更高的速度和更宽的速度范围内进行精密播种。根据插种孔的数量设计了不同的搅拌轮结构。采用离散元素法和计算流体力学相结合的方法分析了气体和种子的运动特性，以确定最佳搅拌轮结构。台架试验结果表明，2、4、6 和 8 个孔插片的最佳播种速度范围分别为 6-9km/h、12-18km/h、18-27km/h 和 24-36km/h。使用 8 孔插条时，最高播种速度达到 36 km/h，漏种率为 2.75%，重复率为 3.76%，合格率为 93.49%。高速离心玉米精密种子计量装置在播种过程中的能耗低于 411.71 千焦/公顷，不到气动种子计量装置每公顷能耗的 9%。此外，播种速度越快，每公顷能耗越低。当播种速度为 36 千米/小时时，能耗为 90.08 千焦/公顷。与气动式种子计量装置相比，高速离心式玉米精密种子计量装置具有更高的播种速度和更低的能耗，可实现高速和清洁生产。

{"title":"Research on high-speed and clean production with a high-speed centrifugal maize precision seed metering device featuring variable hole insert numbers","authors":"Chuan Li , Dongxing Zhang , Li Yang , Tao Cui , Xiantao He , Zhimin Li , Jiaqi Dong , Shulun Xing , Yeyuan Jiang , Jiyuan Liang","doi":"10.1016/j.compag.2024.109620","DOIUrl":"10.1016/j.compag.2024.109620","url":null,"abstract":"<div><div>Traditional pneumatic seed metering devices rely on air pressure for seed filling and carrying, resulting in high energy consumption and limited seeding speed. While centrifugal seed metering devices can achieve high-speed seeding, they have a narrow optimal seeding speed range, making low-speed seeding difficult. In this study, the number of hole inserts in a high-speed centrifugal precision seed metering device for maize was set to 2, 4, 6, and 8, enabling precision seeding at higher speeds and across a broader speed range. Different agitator wheel structures were designed based on the number of hole inserts. The motion characteristics of the gas and seeds were analyzed using a combination of Discrete Element Method and Computational Fluid Dynamics to determine the optimal agitator wheel structure. Bench test results indicated that the optimal seeding speed ranges for 2, 4, 6, and 8 hole inserts were 6–9 km/h, 12–18 km/h, 18–27 km/h, and 24–36 km/h, respectively. With 8 hole inserts, the maximum seeding speed reached 36 km/h, achieving a miss rate of 2.75 %, a repeat rate of 3.76 %, and a qualification rate of 93.49 %. The energy consumption of the high-speed centrifugal maize precision seed metering device during seeding was less than 411.71 kJ/ha, which is less than 9 % of the energy consumed per hectare by pneumatic seed metering devices. Additionally, the higher the seeding speed, the lower the energy consumption per hectare. At a seeding speed of 36 km/h, the energy consumption was 90.08 kJ/ha. Compared to pneumatic seed metering devices, the high-speed centrifugal maize precision seed metering device offers higher seeding speeds and lower energy consumption, enabling high-speed and clean production.</div></div>","PeriodicalId":50627,"journal":{"name":"Computers and Electronics in Agriculture","volume":"227 ","pages":"Article 109620"},"PeriodicalIF":7.7,"publicationDate":"2024-11-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142661764","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Multi objective motion planning of fruit harvesting manipulator based on improved BIT* algorithm 基于改进型 BIT* 算法的水果采摘机械手多目标运动规划

IF 7.7 1区农林科学 Q1 AGRICULTURE, MULTIDISCIPLINARY

Computers and Electronics in Agriculture

Pub Date : 2024-11-12 DOI: 10.1016/j.compag.2024.109567

Peifeng Ma , Aibin Zhu , Yihao Chen , Yao Tu , Han Mao , Jiyuan Song , Xin Wang , Sheng Su , Dangchao Li , Xia Dong

The primary challenge for fruit-harvesting robots in unstructured orchard environments lies in achieving fast and accurate fruit picking while avoiding obstacles like branches. This paper introduces a rapid and efficient multi-objective motion planning method based on the improved BIT* algorithm. Two depth cameras are employed to acquire the locations of both targets and obstacles, and an obstacle map of the harvesting environment is generated using the octree method. For collision detection, a combination of bounding box and grid-based techniques is applied. The proposed bidirectional BIT* (Bi-BIT*) algorithm builds forward and backward trees simultaneously during initialization, alternating searches to reduce the time required for the initial solution. The manipulator’s joint paths are interpolated using a quintic polynomial, and a multi-objective optimization problem is solved to achieve a smooth joint motion trajectory while minimizing energy consumption and pulsation. Both two-dimensional and three-dimensional simulations demonstrate that the Bi-BIT* algorithm consistently outperforms three other algorithms, achieving the highest overall scores. In the harvesting experiment of Scenario 1, the Bi-BIT* algorithm had an average execution time of 7.32 s—36.4% faster than the Informed RRT* algorithm, 19.0% faster than the RRT-Connect algorithm, and 28.7% faster than the BIT* algorithm. Additionally, the Bi-BIT* algorithm achieved a 96% planning success rate and an 84% execution success rate, surpassing the other three algorithms. In Experiment Scenario 2, the Bi-BIT* algorithm had an average execution time of 8.59 s, which is 41.0% faster than the Informed RRT* algorithm, 6.3% faster than the RRT-Connect algorithm, and 19.5% faster than the BIT* algorithm. Furthermore, the Bi-BIT* algorithm demonstrated superior planning and execution success rates of 92% and 88%, respectively, compared to the other algorithms. These experimental results confirm that the proposed multi-objective motion planning method enables the harvesting manipulator to avoid obstacles efficiently and accurately, completing the harvesting task with high performance.

在非结构化果园环境中，水果采摘机器人面临的主要挑战是在避开树枝等障碍物的同时实现快速、准确的水果采摘。本文基于改进的 BIT* 算法，介绍了一种快速高效的多目标运动规划方法。采用两个深度摄像头获取目标和障碍物的位置，并使用八叉树方法生成采摘环境的障碍物地图。在碰撞检测方面，采用了边界框和基于网格的组合技术。所提出的双向 BIT* (Bi-BIT*) 算法在初始化过程中同时建立前向树和后向树，交替搜索以减少初始解所需的时间。使用五次多项式对机械手的关节路径进行插值，并解决多目标优化问题，以实现平滑的关节运动轨迹，同时最大限度地减少能耗和脉动。二维和三维模拟结果表明，Bi-BIT* 算法始终优于其他三种算法，总分最高。在场景 1 的收割实验中，Bi-BIT* 算法的平均执行时间为 7.32 秒，比 Informed RRT* 算法快 36.4%，比 RRT-Connect 算法快 19.0%，比 BIT* 算法快 28.7%。此外，Bi-BIT* 算法的规划成功率达到 96%，执行成功率达到 84%，超过了其他三种算法。在实验方案 2 中，Bi-BIT* 算法的平均执行时间为 8.59 秒，比 Informed RRT* 算法快 41.0%，比 RRT-Connect 算法快 6.3%，比 BIT* 算法快 19.5%。此外，与其他算法相比，Bi-BIT* 算法的规划和执行成功率更高，分别达到 92% 和 88%。这些实验结果证实，所提出的多目标运动规划方法能使收割机械手高效、准确地避开障碍物，高性能地完成收割任务。

{"title":"Multi objective motion planning of fruit harvesting manipulator based on improved BIT* algorithm","authors":"Peifeng Ma , Aibin Zhu , Yihao Chen , Yao Tu , Han Mao , Jiyuan Song , Xin Wang , Sheng Su , Dangchao Li , Xia Dong","doi":"10.1016/j.compag.2024.109567","DOIUrl":"10.1016/j.compag.2024.109567","url":null,"abstract":"<div><div>The primary challenge for fruit-harvesting robots in unstructured orchard environments lies in achieving fast and accurate fruit picking while avoiding obstacles like branches. This paper introduces a rapid and efficient multi-objective motion planning method based on the improved BIT* algorithm. Two depth cameras are employed to acquire the locations of both targets and obstacles, and an obstacle map of the harvesting environment is generated using the octree method. For collision detection, a combination of bounding box and grid-based techniques is applied. The proposed bidirectional BIT* (Bi-BIT*) algorithm builds forward and backward trees simultaneously during initialization, alternating searches to reduce the time required for the initial solution. The manipulator’s joint paths are interpolated using a quintic polynomial, and a multi-objective optimization problem is solved to achieve a smooth joint motion trajectory while minimizing energy consumption and pulsation. Both two-dimensional and three-dimensional simulations demonstrate that the Bi-BIT* algorithm consistently outperforms three other algorithms, achieving the highest overall scores. In the harvesting experiment of Scenario 1, the Bi-BIT* algorithm had an average execution time of 7.32 s—36.4% faster than the Informed RRT* algorithm, 19.0% faster than the RRT-Connect algorithm, and 28.7% faster than the BIT* algorithm. Additionally, the Bi-BIT* algorithm achieved a 96% planning success rate and an 84% execution success rate, surpassing the other three algorithms. In Experiment Scenario 2, the Bi-BIT* algorithm had an average execution time of 8.59 s, which is 41.0% faster than the Informed RRT* algorithm, 6.3% faster than the RRT-Connect algorithm, and 19.5% faster than the BIT* algorithm. Furthermore, the Bi-BIT* algorithm demonstrated superior planning and execution success rates of 92% and 88%, respectively, compared to the other algorithms. These experimental results confirm that the proposed multi-objective motion planning method enables the harvesting manipulator to avoid obstacles efficiently and accurately, completing the harvesting task with high performance.</div></div>","PeriodicalId":50627,"journal":{"name":"Computers and Electronics in Agriculture","volume":"227 ","pages":"Article 109567"},"PeriodicalIF":7.7,"publicationDate":"2024-11-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142658569","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Calibration of mass-spring-damper equivalent systems for real time assessment of the dynamics of trees 校准用于实时评估树木动态的质量-弹簧-阻尼等效系统

IF 7.7 1区农林科学 Q1 AGRICULTURE, MULTIDISCIPLINARY

Computers and Electronics in Agriculture

Pub Date : 2024-11-12 DOI: 10.1016/j.compag.2024.109610

Ernesto Grande , Raffaella Franceschini

In-situ tests and numerical models represent valuable tools for deriving the main dynamic characteristics of trees and for studying their response to dynamic actions. Regarding the numerical models, a key aspect is their calibration. Most procedures available in the literature generally suggest the use of a significant number of instruments (accelerometers placed on both the trunk and branches), which results in high costs and is time-consuming. The aim of this paper is to propose a two-phase approach to calibrate multiple mass-spring-damper systems for studying the dynamics of trees. The proposal aims to support the monitoring and stability assessment of trees through an efficient procedure that combines techniques and methods derived from the field of structural dynamics. Some of these techniques are already used for trees, while others are newly applied in this context. In particular, the experimental data deduced from pull-release tests performed using a single accelerometer placed only on the trunk are assumed as the input data for the approach. The approach is presented in the first part of the paper. In the second part, the approach is implemented in the computer code Matlab to validate it with reference to both numerical models and real tree cases. Finally, a user-friendly graphical application of the approach is developed to make it a practical and expedient tool for researchers and practitioners, allowing real-time evaluation of the dynamics of trees, conducted simultaneously with in-situ tests.

现场试验和数值模型是得出树木主要动态特性和研究其对动态作用响应的重要工具。关于数值模型，一个关键的方面是对其进行校准。文献中的大多数程序一般都建议使用大量仪器（放置在树干和树枝上的加速度计），这不仅成本高，而且耗时。本文旨在提出一种分两个阶段校准多个质量-弹簧-阻尼系统的方法，用于研究树木的动力学。该建议旨在通过结合结构动力学领域的技术和方法的高效程序，为树木的监测和稳定性评估提供支持。其中一些技术已经用于树木，而另一些技术则是新应用于树木。特别是，该方法的输入数据假定是通过仅放置在树干上的单个加速度计进行的拉力释放试验得出的。本文第一部分介绍了该方法。在第二部分中，该方法在计算机代码 Matlab 中实现，并参考数值模型和真实树木案例进行验证。最后，还开发了该方法的用户友好型图形应用程序，使其成为研究人员和从业人员的实用便捷工具，可与现场测试同时进行，对树木的动态进行实时评估。

{"title":"Calibration of mass-spring-damper equivalent systems for real time assessment of the dynamics of trees","authors":"Ernesto Grande , Raffaella Franceschini","doi":"10.1016/j.compag.2024.109610","DOIUrl":"10.1016/j.compag.2024.109610","url":null,"abstract":"<div><div>In-situ tests and numerical models represent valuable tools for deriving the main dynamic characteristics of trees and for studying their response to dynamic actions. Regarding the numerical models, a key aspect is their calibration. Most procedures available in the literature generally suggest the use of a significant number of instruments (accelerometers placed on both the trunk and branches), which results in high costs and is time-consuming. The aim of this paper is to propose a two-phase approach to calibrate multiple mass-spring-damper systems for studying the dynamics of trees. The proposal aims to support the monitoring and stability assessment of trees through an efficient procedure that combines techniques and methods derived from the field of structural dynamics. Some of these techniques are already used for trees, while others are newly applied in this context. In particular, the experimental data deduced from pull-release tests performed using a single accelerometer placed only on the trunk are assumed as the input data for the approach. The approach is presented in the first part of the paper. In the second part, the approach is implemented in the computer code Matlab to validate it with reference to both numerical models and real tree cases. Finally, a user-friendly graphical application of the approach is developed to make it a practical and expedient tool for researchers and practitioners, allowing real-time evaluation of the dynamics of trees, conducted simultaneously with in-situ tests.</div></div>","PeriodicalId":50627,"journal":{"name":"Computers and Electronics in Agriculture","volume":"227 ","pages":"Article 109610"},"PeriodicalIF":7.7,"publicationDate":"2024-11-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142661766","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Non-destructive detection of wheat moisture content with frequency modulated continuous wave system under L and S bands 利用 L 波段和 S 波段下的频率调制连续波系统对小麦水分含量进行无损检测

IF 7.7 1区农林科学 Q1 AGRICULTURE, MULTIDISCIPLINARY

Computers and Electronics in Agriculture

Pub Date : 2024-11-12 DOI: 10.1016/j.compag.2024.109644

Xiaofei Kuang, Zhe Zhu, Jiao Guo, Shiyu Xiang

Wheat moisture content is a critical indicator for evaluating quality. The microwave free space measurement method can achieve nondestructive and efficient measurement of wheat moisture. Regarding microwave detection technology for wheat moisture content, further validation is needed for establishing a prediction model using multi-frequency and full-frequency data within a specific band. Due to the excellent penetration capability of microwaves in the L and S bands, this study explores the potential of utilizing multi-frequency and full-frequency signals in these bands to develop a prediction system for wheat water content. The paper analyzes the relationship between different microwave frequencies, temperatures, moisture contents, and bulk densities on dielectric properties. Temperature, bulk density, and dielectric properties serve as characteristic parameters for the regression model, and a moisture prediction model incorporating single frequency, multi-frequency, and full-frequency data is established. The moisture content detection model integrates three regression methods: Partial Least Squares (PLS), Support Vector Regression (SVR), and Extreme Learning Machine (ELM). Results show that among the nine different prediction models, the SVR model under full-frequency conditions performs the best. The correlation coefficient, root mean square error, and residual prediction bias for moisture prediction on the validation set are 0.9838, 0.3511%, and 6.3245, respectively. To enable online detection of wheat moisture content, a low-cost frequency modulated continuous wave (FMCW) detection system was designed based on the optimal prediction model. Experiments have confirmed that within the moisture content range of 11.35% to 17.79%, the average determination coefficient between the moisture content obtained through drying methods and the measurement results from the FMCW system can reach 0.9493. These endeavors have the potential to provide reliable and cost-effective solutions for precision agriculture applications.

小麦水分含量是评价质量的一个重要指标。微波自由空间测量方法可以实现小麦水分的无损和高效测量。关于小麦水分含量的微波检测技术，需要进一步验证，以便利用特定波段内的多频和全频数据建立预测模型。由于 L 波段和 S 波段的微波具有出色的穿透能力，本研究探讨了利用这些波段的多频和全频信号开发小麦含水量预测系统的潜力。本文分析了不同微波频率、温度、含水量和容重对介电性质的影响。温度、容重和介电性质作为回归模型的特征参数，建立了一个包含单频、多频和全频数据的水分预测模型。含水率检测模型集成了三种回归方法：部分最小二乘法（PLS）、支持向量回归法（SVR）和极限学习机（ELM）。结果表明，在九种不同的预测模型中，全频条件下的 SVR 模型表现最佳。验证集上水分预测的相关系数、均方根误差和剩余预测偏差分别为 0.9838、0.3511% 和 6.3245。为了实现小麦水分含量的在线检测，根据最优预测模型设计了一种低成本的频率调制连续波（FMCW）检测系统。实验证实，在 11.35% 至 17.79% 的水分含量范围内，通过干燥方法获得的水分含量与 FMCW 系统测量结果之间的平均确定系数可达 0.9493。这些努力有可能为精准农业应用提供可靠且经济高效的解决方案。

{"title":"Non-destructive detection of wheat moisture content with frequency modulated continuous wave system under L and S bands","authors":"Xiaofei Kuang, Zhe Zhu, Jiao Guo, Shiyu Xiang","doi":"10.1016/j.compag.2024.109644","DOIUrl":"10.1016/j.compag.2024.109644","url":null,"abstract":"<div><div>Wheat moisture content is a critical indicator for evaluating quality. The microwave free space measurement method can achieve nondestructive and efficient measurement of wheat moisture. Regarding microwave detection technology for wheat moisture content, further validation is needed for establishing a prediction model using multi-frequency and full-frequency data within a specific band. Due to the excellent penetration capability of microwaves in the L and S bands, this study explores the potential of utilizing multi-frequency and full-frequency signals in these bands to develop a prediction system for wheat water content. The paper analyzes the relationship between different microwave frequencies, temperatures, moisture contents, and bulk densities on dielectric properties. Temperature, bulk density, and dielectric properties serve as characteristic parameters for the regression model, and a moisture prediction model incorporating single frequency, multi-frequency, and full-frequency data is established. The moisture content detection model integrates three regression methods: Partial Least Squares (PLS), Support Vector Regression (SVR), and Extreme Learning Machine (ELM). Results show that among the nine different prediction models, the SVR model under full-frequency conditions performs the best. The correlation coefficient, root mean square error, and residual prediction bias for moisture prediction on the validation set are 0.9838, 0.3511%, and 6.3245, respectively. To enable online detection of wheat moisture content, a low-cost frequency modulated continuous wave (FMCW) detection system was designed based on the optimal prediction model. Experiments have confirmed that within the moisture content range of 11.35% to 17.79%, the average determination coefficient between the moisture content obtained through drying methods and the measurement results from the FMCW system can reach 0.9493. These endeavors have the potential to provide reliable and cost-effective solutions for precision agriculture applications.</div></div>","PeriodicalId":50627,"journal":{"name":"Computers and Electronics in Agriculture","volume":"227 ","pages":"Article 109644"},"PeriodicalIF":7.7,"publicationDate":"2024-11-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142661776","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Tomato maturity detection based on bioelectrical impedance spectroscopy 基于生物电阻抗光谱的番茄成熟度检测

IF 7.7 1区农林科学 Q1 AGRICULTURE, MULTIDISCIPLINARY

Computers and Electronics in Agriculture

Pub Date : 2024-11-12 DOI: 10.1016/j.compag.2024.109553

Zhang Yongnian , Chen Yinhe , Bao Yihua , Wang Xiaochan , Xian Jieyu

This paper proposes a method for detecting tomato maturity to address issues related to post-harvest storage and transportation. The method utilizes bioimpedance spectroscopy to study tomato maturity, construct the Double-R-Cole equivalent circuit model, and obtain electrical parameters through fitting the Levenberg-Marquardt optimization algorithm. We analyze the change rule of electrical parameters in different ripening periods, use Fisher’s discriminant to reduce the dimensionality of features such as biological variables, fitted electrical parameters, and storage days of tomato, and combine the advantages of support vector machine and random forest to classify the input features. The classification algorithm utilizes the gorilla troop optimization algorithm to address issues with traditional iterative algorithms, such as difficulty assigning initial values and susceptibility to local optima. The study finds that: the Levenberg-Marquardt algorithm fitted an R^2 mean value of 0.997 and the significance level of p < 0.001 between the two constant-phase components of the fitted electrical parameters and the number of storage days proved that the established Double-R-Cole model could effectively characterize the postharvest situation of tomato; the Fisher’s discriminant based SVM-RF-GTO’s maturity classification algorithm achieves 97.26 % effectiveness in tomato maturity classification. This research provides valuable insights for tomato postharvest storage and transportation.

本文提出了一种检测番茄成熟度的方法，以解决与收获后贮藏和运输相关的问题。该方法利用生物阻抗光谱研究番茄成熟度，构建 Double-R-Cole 等效电路模型，并通过 Levenberg-Marquardt 优化算法拟合获得电参数。分析不同成熟期电参数的变化规律，利用费雪判别法降低番茄生物变量、拟合电参数、贮藏天数等特征的维度，并结合支持向量机和随机森林的优点对输入特征进行分类。分类算法采用了猩猩部队优化算法，以解决传统迭代算法存在的问题，如初始值分配困难和易出现局部最优等。研究发现：Levenberg-Marquardt 算法拟合的 R^2 均值为 0.997，拟合电参数的两个常相分量与贮藏天数之间的显著性水平为 p < 0.001，证明建立的 Double-R-Cole 模型能有效表征番茄采后情况；基于 Fisher 判别的 SVM-RF-GTO 成熟度分类算法在番茄成熟度分类中的有效性达到 97.26%。这项研究为番茄采后储运提供了有价值的见解。

{"title":"Tomato maturity detection based on bioelectrical impedance spectroscopy","authors":"Zhang Yongnian , Chen Yinhe , Bao Yihua , Wang Xiaochan , Xian Jieyu","doi":"10.1016/j.compag.2024.109553","DOIUrl":"10.1016/j.compag.2024.109553","url":null,"abstract":"<div><div>This paper proposes a method for detecting tomato maturity to address issues related to post-harvest storage and transportation. The method utilizes bioimpedance spectroscopy to study tomato maturity, construct the Double-R-Cole equivalent circuit model, and obtain electrical parameters through fitting the Levenberg-Marquardt optimization algorithm. We analyze the change rule of electrical parameters in different ripening periods, use Fisher’s discriminant to reduce the dimensionality of features such as biological variables, fitted electrical parameters, and storage days of tomato, and combine the advantages of support vector machine and random forest to classify the input features. The classification algorithm utilizes the gorilla troop optimization algorithm to address issues with traditional iterative algorithms, such as difficulty assigning initial values and susceptibility to local optima. The study finds that: the Levenberg-Marquardt algorithm fitted an R^2 mean value of 0.997 and the significance level of <em>p</em> < 0.001 between the two constant-phase components of the fitted electrical parameters and the number of storage days proved that the established Double-R-Cole model could effectively characterize the postharvest situation of tomato; the Fisher’s discriminant based SVM-RF-GTO’s maturity classification algorithm achieves 97.26 % effectiveness in tomato maturity classification. This research provides valuable insights for tomato postharvest storage and transportation.</div></div>","PeriodicalId":50627,"journal":{"name":"Computers and Electronics in Agriculture","volume":"227 ","pages":"Article 109553"},"PeriodicalIF":7.7,"publicationDate":"2024-11-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142661779","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Developments in deep learning approaches for apple leaf Alternaria disease identification: A review 深度学习方法在苹果叶片交替侵染病识别中的发展：综述

IF 7.7 1区农林科学 Q1 AGRICULTURE, MULTIDISCIPLINARY

Computers and Electronics in Agriculture

Pub Date : 2024-11-11 DOI: 10.1016/j.compag.2024.109593

Mansoor Ahmad Kirmani, Yasir Afaq

Apple tree leaf diseases (ATLDs) can be accurately identified and addressed early to prevent the diseases from spreading, minimize the need for chemical pesticides and fertilizers, increase apple quality and production, and preserve the healthy growth of apple varieties. To overcome such challenges, different Deep Learning (DL) approaches have been developed to early detect apple leaf diseases. In this paper, the data from 2010 to 2024 has been taken for analysis, and it has been observed that many of the researchers have utilized different types of datasets for disease detection. Moreover, Deep Learning (DL) and Machine Learning (ML) have been mostly utilized for the detection and identification of apple leaf Alternaria diseases. It has also been observed from the previous work that Support Vector Machines (SVM), Random Forests (RF), XGBoost, and many more are the most common approaches utilized by the researchers. On the other hand, DenseNet, MobileNet, Convolutional Neural Network (CNN), and Vision Transformer are the deep learning approaches utilized by the researchers. Furthermore, we have also given a brief analysis of each approach along with a comparative analysis such as lightweight CNNs and Attention-based mechanisms, Transfer Learning (TL), Localization techniques, Vision Transformer (ViT), and Severity estimation techniques. Emphasizing their methods, datasets, performance metrics, and real-world applications. This study explores the proposed models’ approaches, feature selection and extraction techniques, data capturing conditions, accuracy, types of datasets used in the experiments, and their resources. Our research findings indicate that although DL approaches have significant potential for improving disease management in agriculture. There is a crucial need for a more scalable, robust, and flexible solution to handle numerous agricultural conditions and disease complexities. By methodically and comprehensively analyzing the collected data, this study aims to facilitate valuable resources for researchers aiming to design, develop, and implement DL-based systems for apple leaf disease detection and identification, ultimately contributing to sustainable agriculture and improved food security.

苹果树叶部病害（ATLDs）可以准确识别并及早解决，以防止病害蔓延，最大限度地减少对化学农药和化肥的需求，提高苹果质量和产量，保护苹果品种的健康生长。为了克服这些挑战，人们开发了不同的深度学习（DL）方法来早期检测苹果叶片病害。本文分析了 2010 年至 2024 年的数据，发现许多研究人员利用不同类型的数据集进行疾病检测。此外，深度学习（DL）和机器学习（ML）也被广泛应用于苹果叶片白粉病的检测和识别。从以往的工作中还可以看出，支持向量机（SVM）、随机森林（RF）、XGBoost 等是研究人员最常用的方法。另一方面，DenseNet、MobileNet、卷积神经网络（CNN）和 Vision Transformer 是研究人员使用的深度学习方法。此外，我们还对每种方法进行了简要分析和比较分析，如轻量级 CNN 和基于注意力的机制、迁移学习（TL）、定位技术、视觉转换器（ViT）和严重性估计技术。研究强调了它们的方法、数据集、性能指标和实际应用。本研究探讨了拟议模型的方法、特征选择和提取技术、数据捕获条件、准确性、实验中使用的数据集类型及其资源。我们的研究结果表明，尽管 DL 方法在改善农业病害管理方面具有巨大潜力，但在农业病害管理方面还存在许多问题。但亟需一种更具可扩展性、稳健性和灵活性的解决方案来处理众多农业条件和复杂的疾病。通过有条不紊地全面分析收集到的数据，本研究旨在为旨在设计、开发和实施基于 DL 的苹果叶病检测和识别系统的研究人员提供宝贵的资源，最终为可持续农业和提高粮食安全做出贡献。

{"title":"Developments in deep learning approaches for apple leaf Alternaria disease identification: A review","authors":"Mansoor Ahmad Kirmani, Yasir Afaq","doi":"10.1016/j.compag.2024.109593","DOIUrl":"10.1016/j.compag.2024.109593","url":null,"abstract":"<div><div>Apple tree leaf diseases (ATLDs) can be accurately identified and addressed early to prevent the diseases from spreading, minimize the need for chemical pesticides and fertilizers, increase apple quality and production, and preserve the healthy growth of apple varieties. To overcome such challenges, different Deep Learning (DL) approaches have been developed to early detect apple leaf diseases. In this paper, the data from 2010 to 2024 has been taken for analysis, and it has been observed that many of the researchers have utilized different types of datasets for disease detection. Moreover, Deep Learning (DL) and Machine Learning (ML) have been mostly utilized for the detection and identification of apple leaf Alternaria diseases. It has also been observed from the previous work that Support Vector Machines (SVM), Random Forests (RF), XGBoost, and many more are the most common approaches utilized by the researchers. On the other hand, DenseNet, MobileNet, Convolutional Neural Network (CNN), and Vision Transformer are the deep learning approaches utilized by the researchers. Furthermore, we have also given a brief analysis of each approach along with a comparative analysis such as lightweight CNNs and Attention-based mechanisms, Transfer Learning (TL), Localization techniques, Vision Transformer (ViT), and Severity estimation techniques. Emphasizing their methods, datasets, performance metrics, and real-world applications. This study explores the proposed models’ approaches, feature selection and extraction techniques, data capturing conditions, accuracy, types of datasets used in the experiments, and their resources. Our research findings indicate that although DL approaches have significant potential for improving disease management in agriculture. There is a crucial need for a more scalable, robust, and flexible solution to handle numerous agricultural conditions and disease complexities. By methodically and comprehensively analyzing the collected data, this study aims to facilitate valuable resources for researchers aiming to design, develop, and implement DL-based systems for apple leaf disease detection and identification, ultimately contributing to sustainable agriculture and improved food security.</div></div>","PeriodicalId":50627,"journal":{"name":"Computers and Electronics in Agriculture","volume":"227 ","pages":"Article 109593"},"PeriodicalIF":7.7,"publicationDate":"2024-11-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142661364","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Multi-model fusion method for predicting CO2 concentration in greenhouse tomatoes 预测温室番茄二氧化碳浓度的多模型融合方法

IF 7.7 1区农林科学 Q1 AGRICULTURE, MULTIDISCIPLINARY

Computers and Electronics in Agriculture

Pub Date : 2024-11-10 DOI: 10.1016/j.compag.2024.109623

Jianjun Guo , Beibei Zhang , Lijun Lin , Yudian Xu , Piao Zhou , Shangwen Luo , Yuhan Zhuo , Jingyu Ji , Zhijie Luo , Shahbaz Gul Hassan

With the rapid development of greenhouse agriculture, accurate prediction of environmental parameters such as temperature, humidity, and carbon dioxide concentration is crucial for optimal crop growth. Traditional forecasting models struggle with the nonlinear and complex nature of greenhouse data, leading to challenges in model robustness. This study addresses these issues by proposing a multi-model fusion strategy for predicting CO2 concentration in greenhouse tomatoes. The proposed method integrates wavelet denoising (WT), variational mode decomposition (VMD), and long short-term memory networks (LSTM). This innovative nonlinear ensemble model effectively extracts key time series features and removes noise, while an introduced attention mechanism enhances the model’s focus on essential time steps, improving prediction accuracy. Experimental results demonstrate that the multi-model fusion approach significantly outperforms single models in terms of accuracy and stability, achieving mean absolute error (MAE) and root mean square error (RMSE) of 0.0117 and 0.0194, respectively. The proposed method offers significant advantages for CO2 prediction in greenhouse crops, providing a theoretical basis and technical support for optimizing and controlling greenhouse parameters. This contributes to the advancement of smart agriculture by offering an efficient environmental monitoring and prediction tool. Additionally, the study presents new ideas and technical solutions for addressing similar agricultural environment prediction challenges, optimizing greenhouse environment control strategies, and improving crop production efficiency.

随着温室农业的快速发展，准确预测温度、湿度和二氧化碳浓度等环境参数对作物的最佳生长至关重要。传统的预测模型难以应对温室数据的非线性和复杂性，导致模型的鲁棒性面临挑战。本研究针对这些问题，提出了一种预测温室番茄二氧化碳浓度的多模型融合策略。所提出的方法整合了小波去噪 (WT)、变模分解 (VMD) 和长短期记忆网络 (LSTM)。这种创新的非线性集合模型能有效提取关键的时间序列特征并去除噪声，同时引入的注意力机制能增强模型对重要时间步骤的关注，从而提高预测精度。实验结果表明，多模型融合方法在准确性和稳定性方面明显优于单一模型，平均绝对误差（MAE）和均方根误差（RMSE）分别为 0.0117 和 0.0194。所提出的方法在温室作物二氧化碳预测方面具有显著优势，为优化和控制温室参数提供了理论依据和技术支持。通过提供高效的环境监测和预测工具，该方法有助于推动智能农业的发展。此外，该研究还为解决类似的农业环境预测难题、优化温室环境控制策略和提高作物生产效率提出了新的思路和技术解决方案。

{"title":"Multi-model fusion method for predicting CO2 concentration in greenhouse tomatoes","authors":"Jianjun Guo , Beibei Zhang , Lijun Lin , Yudian Xu , Piao Zhou , Shangwen Luo , Yuhan Zhuo , Jingyu Ji , Zhijie Luo , Shahbaz Gul Hassan","doi":"10.1016/j.compag.2024.109623","DOIUrl":"10.1016/j.compag.2024.109623","url":null,"abstract":"<div><div>With the rapid development of greenhouse agriculture, accurate prediction of environmental parameters such as temperature, humidity, and carbon dioxide concentration is crucial for optimal crop growth. Traditional forecasting models struggle with the nonlinear and complex nature of greenhouse data, leading to challenges in model robustness. This study addresses these issues by proposing a multi-model fusion strategy for predicting CO2 concentration in greenhouse tomatoes. The proposed method integrates wavelet denoising (WT), variational mode decomposition (VMD), and long short-term memory networks (LSTM). This innovative nonlinear ensemble model effectively extracts key time series features and removes noise, while an introduced attention mechanism enhances the model’s focus on essential time steps, improving prediction accuracy. Experimental results demonstrate that the multi-model fusion approach significantly outperforms single models in terms of accuracy and stability, achieving mean absolute error (MAE) and root mean square error (RMSE) of 0.0117 and 0.0194, respectively. The proposed method offers significant advantages for CO2 prediction in greenhouse crops, providing a theoretical basis and technical support for optimizing and controlling greenhouse parameters. This contributes to the advancement of smart agriculture by offering an efficient environmental monitoring and prediction tool. Additionally, the study presents new ideas and technical solutions for addressing similar agricultural environment prediction challenges, optimizing greenhouse environment control strategies, and improving crop production efficiency.</div></div>","PeriodicalId":50627,"journal":{"name":"Computers and Electronics in Agriculture","volume":"227 ","pages":"Article 109623"},"PeriodicalIF":7.7,"publicationDate":"2024-11-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142661767","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Spatial interpolators for Delineating management zones to mitigate Mucuna pruriens in sugarcane plantations in the Eastern Amazon 用于划定管理区的空间插值器，以减少亚马逊河东部甘蔗种植园中的金丝桃危害

IF 7.7 1区农林科学 Q1 AGRICULTURE, MULTIDISCIPLINARY

Computers and Electronics in Agriculture

Pub Date : 2024-11-10 DOI: 10.1016/j.compag.2024.109615

Luiz Antonio Soares Cardoso , Paulo Roberto Silva Farias , João Almiro Corrêa Soares , Carlos Rodrigo Tanajura Caldeira , Fábio Júnior de Oliveira

Unmanned Aerial Vehicles (UAVs) have emerged as essential tools in precision agriculture, employing aerial photogrammetry concepts to aid producers in various decision-making processes. This study evauates different spatial interpolators to define management zones in sugarcane fields, aiming to control potential infestations by Mucuna pruriens. We collected images using the EbeeSQ UAV, equipped with a multispectral sensor, and calculated vegetation indices, including NDVI, SAVI, NDRE, and GNDVI. Analysis revealed that GNDVI yielded the most favorable results, with a mean value of 0.304 and a coefficient of variation of 11.747 %. Using regular and random sampling grids, we applied Ordinary Kriging (OK) and Support Vector Machine (SVM) interpolators to assess spatial variability across 13 survey zones. The results indicated a Degree of Spatial Dependence averaging 57.197 % and a Moran Index of 0.609, confirming moderate spatial dependence. Cross-validation showed that OK with random sampling outperformed other methods, achieving a Root Mean Square Error (RMSE) of 0.064 and a coefficient of determination (r²) averaging 0.347. Furthermore, the relationship between the Fuzzy Performance Index (averaging 0.069) and Normalized Classification Entropy (averaging 0.077) enabled the creation of management zone maps. These maps effectively identify distinct classes within the study areas, enhancing decision-making for producers in managing velvet bean weed during critical developmental phases.

无人驾驶飞行器（UAV）已成为精准农业的重要工具，它采用航空摄影测量概念来帮助生产者进行各种决策。本研究利用不同的空间插值法来确定甘蔗田的管理区域，目的是控制潜在的金丝楠木侵扰。我们使用配备多光谱传感器的 EbeeSQ 无人机采集图像，并计算植被指数，包括 NDVI、SAVI、NDRE 和 GNDVI。分析表明，GNDVI 的结果最理想，平均值为 0.304，变异系数为 11.747%。我们使用规则和随机抽样网格，应用普通克里金（OK）和支持向量机（SVM）插值器评估了 13 个调查区的空间变异性。结果表明，空间依赖度平均为 57.197%，莫兰指数为 0.609，证实了中等程度的空间依赖性。交叉验证结果表明，采用随机抽样的 OK 方法优于其他方法，其均方根误差（RMSE）为 0.064，判定系数（r2）平均为 0.347。此外，模糊性能指数（平均为 0.069）和归一化分类熵（平均为 0.077）之间的关系使得管理区域图得以绘制。这些地图有效地确定了研究区域内的不同等级，从而提高了生产者在关键生长阶段管理绒豆杂草的决策能力。

{"title":"Spatial interpolators for Delineating management zones to mitigate Mucuna pruriens in sugarcane plantations in the Eastern Amazon","authors":"Luiz Antonio Soares Cardoso , Paulo Roberto Silva Farias , João Almiro Corrêa Soares , Carlos Rodrigo Tanajura Caldeira , Fábio Júnior de Oliveira","doi":"10.1016/j.compag.2024.109615","DOIUrl":"10.1016/j.compag.2024.109615","url":null,"abstract":"<div><div>Unmanned Aerial Vehicles (UAVs) have emerged as essential tools in precision agriculture, employing aerial photogrammetry concepts to aid producers in various decision-making processes. This study evauates different spatial interpolators to define management zones in sugarcane fields, aiming to control potential infestations by Mucuna pruriens. We collected images using the EbeeSQ UAV, equipped with a multispectral sensor, and calculated vegetation indices, including NDVI, SAVI, NDRE, and GNDVI. Analysis revealed that GNDVI yielded the most favorable results, with a mean value of 0.304 and a coefficient of variation of 11.747 %. Using regular and random sampling grids, we applied Ordinary Kriging (OK) and Support Vector Machine (SVM) interpolators to assess spatial variability across 13 survey zones. The results indicated a Degree of Spatial Dependence averaging 57.197 % and a Moran Index of 0.609, confirming moderate spatial dependence. Cross-validation showed that OK with random sampling outperformed other methods, achieving a Root Mean Square Error (RMSE) of 0.064 and a coefficient of determination (r<sup>2</sup>) averaging 0.347. Furthermore, the relationship between the Fuzzy Performance Index (averaging 0.069) and Normalized Classification Entropy (averaging 0.077) enabled the creation of management zone maps. These maps effectively identify distinct classes within the study areas, enhancing decision-making for producers in managing velvet bean weed during critical developmental phases.</div></div>","PeriodicalId":50627,"journal":{"name":"Computers and Electronics in Agriculture","volume":"227 ","pages":"Article 109615"},"PeriodicalIF":7.7,"publicationDate":"2024-11-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142661362","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

StraTracker: A dynamic counting method for growing strawberries based on multi-target tracking StraTracker：基于多目标跟踪的草莓生长动态计数法

IF 7.7 1区农林科学 Q1 AGRICULTURE, MULTIDISCIPLINARY

Computers and Electronics in Agriculture

Pub Date : 2024-11-09 DOI: 10.1016/j.compag.2024.109564

Qilin An, Yongzhi Cui, Wenyu Tong, Yangchun Liu, Bo Zhao, Liguo Wei

Accurately counting fruit in orchards is a critical step for effective digital farming management. However, the variability in fruit size, overlapping shadows, and light interference present significant challenges to applying computer vision during the strawberry growth phase. To address these challenges, we propose StraTracker, a multi-object tracking (MOT) algorithm specifically designed to identify and count strawberries at various growth stages. StraTracker transforms the counting task into a frame-by-frame tracking problem, integrating both motion and appearance features. The algorithm is composed of three key components: a strawberry detector based on YOLOv8n, a feature association module, and a dual-area counting (DC) module. First, the strawberry detector accurately recognizes five growth stages, achieving an average accuracy of 91.93 % at 38.3 FPS. Next, the feature association module, incorporating the Feature Slicing Attention (FSA) and Adaptive Kalman Filtering (AKF) modules, mitigates issues such as light interference, impractical tracking frames, and ID switching (IDs). As a result, StraTracker achieves a Multi-Object Tracking Accuracy (MOTA) of 83.28 % and a Higher-Order Tracking Accuracy (HOTA) of 77.26 %, with only 259 IDs, outperforming existing baseline models. Finally, the DC module categorizes fruit counts based on the unique IDs assigned during tracking. The algorithm’s coefficient of determination (

R^{2}

= 0.91) and GEH of 2.33 indicate a strong correlation between predicted and actual counts. In conclusion, StraTracker offers a promising solution for farmers to optimize planting strategies and develop more precise harvesting plans.

对果园中的果实进行精确计数是实现有效数字农业管理的关键一步。然而，果实大小的可变性、阴影重叠和光线干扰给在草莓生长阶段应用计算机视觉带来了巨大挑战。为了应对这些挑战，我们提出了 StraTracker，这是一种多目标跟踪 (MOT) 算法，专门用于识别和计数处于不同生长阶段的草莓。StraTracker 将计数任务转化为逐帧跟踪问题，同时整合了运动和外观特征。该算法由三个关键部分组成：基于 YOLOv8n 的草莓检测器、特征关联模块和双区域计数（DC）模块。首先，草莓检测器能准确识别五个生长阶段，在 38.3 FPS 下达到 91.93% 的平均准确率。接着，特征关联模块结合了特征切分注意（FSA）和自适应卡尔曼滤波（AKF）模块，缓解了光线干扰、不切实际的跟踪帧和 ID 切换（ID）等问题。因此，StraTracker 的多目标跟踪准确率 (MOTA) 达到 83.28%，高阶跟踪准确率 (HOTA) 达到 77.26%，ID 数量仅为 259 个，优于现有的基线模型。最后，DC 模块根据跟踪过程中分配的唯一 ID 对水果数量进行分类。该算法的判定系数（R2 = 0.91）和 2.33 的 GEH 表明，预测计数与实际计数之间具有很强的相关性。总之，StraTracker 为农民优化种植策略和制定更精确的收获计划提供了一个前景广阔的解决方案。

{"title":"StraTracker: A dynamic counting method for growing strawberries based on multi-target tracking","authors":"Qilin An, Yongzhi Cui, Wenyu Tong, Yangchun Liu, Bo Zhao, Liguo Wei","doi":"10.1016/j.compag.2024.109564","DOIUrl":"10.1016/j.compag.2024.109564","url":null,"abstract":"<div><div>Accurately counting fruit in orchards is a critical step for effective digital farming management. However, the variability in fruit size, overlapping shadows, and light interference present significant challenges to applying computer vision during the strawberry growth phase. To address these challenges, we propose StraTracker, a multi-object tracking (MOT) algorithm specifically designed to identify and count strawberries at various growth stages. StraTracker transforms the counting task into a frame-by-frame tracking problem, integrating both motion and appearance features. The algorithm is composed of three key components: a strawberry detector based on YOLOv8n, a feature association module, and a dual-area counting (DC) module. First, the strawberry detector accurately recognizes five growth stages, achieving an average accuracy of 91.93 % at 38.3 FPS. Next, the feature association module, incorporating the Feature Slicing Attention (FSA) and Adaptive Kalman Filtering (AKF) modules, mitigates issues such as light interference, impractical tracking frames, and ID switching (IDs). As a result, StraTracker achieves a Multi-Object Tracking Accuracy (MOTA) of 83.28 % and a Higher-Order Tracking Accuracy (HOTA) of 77.26 %, with only 259 IDs, outperforming existing baseline models. Finally, the DC module categorizes fruit counts based on the unique IDs assigned during tracking. The algorithm’s coefficient of determination (<span><math><mrow><msup><mrow><mi>R</mi></mrow><mn>2</mn></msup></mrow></math></span> = 0.91) and GEH of 2.33 indicate a strong correlation between predicted and actual counts. In conclusion, StraTracker offers a promising solution for farmers to optimize planting strategies and develop more precise harvesting plans.</div></div>","PeriodicalId":50627,"journal":{"name":"Computers and Electronics in Agriculture","volume":"227 ","pages":"Article 109564"},"PeriodicalIF":7.7,"publicationDate":"2024-11-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142661773","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Convolutional Neural Networks accurately predict soil matric potential from soil, weather, and satellite-derived data 卷积神经网络从土壤、天气和卫星数据中准确预测土壤母质电位

IF 7.7 1区农林科学 Q1 AGRICULTURE, MULTIDISCIPLINARY

Computers and Electronics in Agriculture

Pub Date : 2024-11-09 DOI: 10.1016/j.compag.2024.109597

Carlos Ballester, John Hornbuckle, Brenno Tondato, Rodrigo Filev-Maia

Being able to predict soil moisture dynamics offers water managers the possibility to better plan irrigation events and prevent soil moisture deficits from reaching levels that reduce crop production. Machine learning (ML) model predictions can potentially assist farmers in managing irrigation water more efficiently. In this study, we aimed to assess the accuracy of a set of ML models in predicting soil matric potential seven days ahead in gravity-surface irrigated cotton paddocks and evaluate the models’ performance for longer term predictions (14 days). The ML models used past soil moisture, weather, and satellite-derived crop-related data as features for the input parameters. Input data were structured in tuples that were organised following a 20-day ‘window’ approach that ‘slid’ one position forward after each training round. A convolutional neural network (CNN) model outperformed a Long Short-Term Memory, Dense Multilayer Perceptron, and Linear Regression model, the latter of which produced the least accurate predictions. The accuracy of the soil matric potential predictions with the CNN model was stable over time (R² ≥ 0.92 and root mean square deviation ≤ 7.5 kPa). However, less accurate predictions were obtained for a short period after emergence and at crop senescence. This study demonstrates the feasibility of producing accurate predictions of soil matric potential in cotton fields at 0.20 m soil depth with a CNN model, which can be integrated into irrigation decision support systems.

能够预测土壤水分动态为水资源管理者更好地规划灌溉活动和防止土壤水分不足达到降低作物产量的水平提供了可能。机器学习（ML）模型预测有可能帮助农民更有效地管理灌溉用水。在这项研究中，我们旨在评估一组 ML 模型在重力地面灌溉棉田中提前 7 天预测土壤水分潜力的准确性，并评估模型在长期预测（14 天）中的性能。ML 模型使用过去的土壤水分、天气和卫星作物相关数据作为输入参数特征。输入数据以元组为结构，按照 20 天的 "窗口 "方法进行组织，每轮训练后向前 "滑动 "一个位置。卷积神经网络（CNN）模型优于长短期记忆、密集多层感知器和线性回归模型，后者的预测准确度最低。使用 CNN 模型预测土壤墒情的准确度在一段时间内保持稳定（R2 ≥ 0.92，均方根偏差小于 7.5 千帕）。不过，对出苗后短时间内和作物衰老期的预测精度较低。这项研究证明了利用 CNN 模型准确预测棉田 0.20 米土壤深度土壤母质势的可行性，该模型可集成到灌溉决策支持系统中。

{"title":"Convolutional Neural Networks accurately predict soil matric potential from soil, weather, and satellite-derived data","authors":"Carlos Ballester, John Hornbuckle, Brenno Tondato, Rodrigo Filev-Maia","doi":"10.1016/j.compag.2024.109597","DOIUrl":"10.1016/j.compag.2024.109597","url":null,"abstract":"<div><div>Being able to predict soil moisture dynamics offers water managers the possibility to better plan irrigation events and prevent soil moisture deficits from reaching levels that reduce crop production. Machine learning (ML) model predictions can potentially assist farmers in managing irrigation water more efficiently. In this study, we aimed to assess the accuracy of a set of ML models in predicting soil matric potential seven days ahead in gravity-surface irrigated cotton paddocks and evaluate the models’ performance for longer term predictions (14 days). The ML models used past soil moisture, weather, and satellite-derived crop-related data as features for the input parameters. Input data were structured in tuples that were organised following a 20-day ‘window’ approach that ‘slid’ one position forward after each training round. A convolutional neural network (CNN) model outperformed a Long Short-Term Memory, Dense Multilayer Perceptron, and Linear Regression model, the latter of which produced the least accurate predictions. The accuracy of the soil matric potential predictions with the CNN model was stable over time (R<sup>2</sup> ≥ 0.92 and root mean square deviation ≤ 7.5 kPa). However, less accurate predictions were obtained for a short period after emergence and at crop senescence. This study demonstrates the feasibility of producing accurate predictions of soil matric potential in cotton fields at 0.20 m soil depth with a CNN model, which can be integrated into irrigation decision support systems.</div></div>","PeriodicalId":50627,"journal":{"name":"Computers and Electronics in Agriculture","volume":"227 ","pages":"Article 109597"},"PeriodicalIF":7.7,"publicationDate":"2024-11-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142661363","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0