Information Processing in Agriculture最新文献_第2页

YOLOv8-DDS: A lightweight model based on pruning and distillation for early detection of root mold in barley seedling YOLOv8-DDS：基于修剪和蒸馏的大麦幼苗根霉早期检测轻量级模型

IF 7.4 Q1 AGRICULTURE, MULTIDISCIPLINARY

Information Processing in Agriculture

Pub Date : 2025-12-01 Epub Date: 2025-07-23 DOI: 10.1016/j.inpa.2025.07.004

Huang Junjie , Ma Zheng , Wu Yuzhu , Bao Yujian , Wang Yizhe , Su Zhongbin , Guo Lifeng

Root mold proliferation presents a significant challenge in the industrial production of hydroponic barley seedlings. The small size, inconspicuous coloration, and indiscernible image of early mold regions pose new demands on detection accuracy. This study constructed a dataset of root mold in barley seedlings throughout their growth cycle and proposed the YOLOv8n-DDS detection model to integrate a lightweight detection model into a three-dimensional cyclic cultivation system. The model incorporates the dynamic sample (DySample) operator, combines deformable ConvNets v2 (DCNv2) with C2f, and reconstructs the detection head using seam carving (SEAM) technology, which enhances its capability to extract multi-scale, minute features of early-stage root mold in barley. To improve the model’s performance on edge-embedded devices, this study employed layer-wise adaptive magnitude pruning and channel-wise knowledge distillation methods, thereby significantly reducing the model’s parameter count and computational load. The pruned and distilled model was subsequently deployed on the Jetson Nano platform for validation. Results indicate that the YOLOv8n-DDS model outperformed the baseline model in terms of precision, recall, and mAP50 by 2.4 %, 5.6 %, and 2.2 %, respectively. The parameter count was reduced by 23.8 %, and the computational complexity (Giga floating-point operators per second) was optimized by 14.8 %. Additionally, the detection latency on resource-constrained embedded devices was further reduced by 25.8 % with TensorRT acceleration. The proposed root mold detection model is lightweight and contributes to the intelligent and technological integration of the industrial production process for high-quality barley seedling forage.

根霉菌增殖是水培大麦幼苗工业化生产中的一个重大挑战。模具早期区域的体积小、颜色不明显、图像难以识别等特点对检测精度提出了新的要求。本研究构建了大麦幼苗整个生长周期的根霉菌数据集，提出了YOLOv8n-DDS检测模型，将轻量化检测模型集成到三维循环栽培系统中。该模型结合动态样本算子（DySample），结合变形卷积神经网络v2 （DCNv2）和C2f算法，利用缝雕刻（seam）技术重构检测头，增强了提取大麦早期根霉多尺度、微小特征的能力。为了提高模型在边缘嵌入式设备上的性能，本研究采用了分层自适应幅度修剪和信道知识升华方法，从而显著减少了模型的参数计数和计算负荷。随后将修剪和提炼的模型部署在Jetson Nano平台上进行验证。结果表明，YOLOv8n-DDS模型在准确率、召回率和mAP50方面分别优于基线模型2.4%、5.6%和2.2%。参数数量减少了23.8%，计算复杂度（每秒千兆浮点运算）优化了14.8%。此外，在资源受限的嵌入式设备上，通过TensorRT加速，检测延迟进一步降低了25.8%。提出的根霉菌检测模型重量轻，有助于高品质大麦苗木饲料工业生产过程的智能化和技术化。

{"title":"YOLOv8-DDS: A lightweight model based on pruning and distillation for early detection of root mold in barley seedling","authors":"Huang Junjie , Ma Zheng , Wu Yuzhu , Bao Yujian , Wang Yizhe , Su Zhongbin , Guo Lifeng","doi":"10.1016/j.inpa.2025.07.004","DOIUrl":"10.1016/j.inpa.2025.07.004","url":null,"abstract":"<div><div>Root mold proliferation presents a significant challenge in the industrial production of hydroponic barley seedlings. The small size, inconspicuous coloration, and indiscernible image of early mold regions pose new demands on detection accuracy. This study constructed a dataset of root mold in barley seedlings throughout their growth cycle and proposed the YOLOv8n-DDS detection model to integrate a lightweight detection model into a three-dimensional cyclic cultivation system. The model incorporates the dynamic sample (DySample) operator, combines deformable ConvNets v2 (DCNv2) with C2f, and reconstructs the detection head using seam carving (SEAM) technology, which enhances its capability to extract multi-scale, minute features of early-stage root mold in barley. To improve the model’s performance on edge-embedded devices, this study employed layer-wise adaptive magnitude pruning and channel-wise knowledge distillation methods, thereby significantly reducing the model’s parameter count and computational load. The pruned and distilled model was subsequently deployed on the Jetson Nano platform for validation. Results indicate that the YOLOv8n-DDS model outperformed the baseline model in terms of precision, recall, and mAP50 by 2.4 %, 5.6 %, and 2.2 %, respectively. The parameter count was reduced by 23.8 %, and the computational complexity (Giga floating-point operators per second) was optimized by 14.8 %. Additionally, the detection latency on resource-constrained embedded devices was further reduced by 25.8 % with TensorRT acceleration. The proposed root mold detection model is lightweight and contributes to the intelligent and technological integration of the industrial production process for high-quality barley seedling forage.</div></div>","PeriodicalId":53443,"journal":{"name":"Information Processing in Agriculture","volume":"12 4","pages":"Pages 581-594"},"PeriodicalIF":7.4,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145697906","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Leveraging close-range UAV phenotyping and GWAS for enhanced understanding of slash pine growth dynamics 利用近距离无人机表型和GWAS来增强对湿地松生长动态的理解

IF 7.4 Q1 AGRICULTURE, MULTIDISCIPLINARY

Information Processing in Agriculture

Pub Date : 2025-12-01 Epub Date: 2025-07-24 DOI: 10.1016/j.inpa.2025.07.002

Xianyin Ding , Pieter B. Pelser , Cong Xu , Ilga Porth , Mingming Cui , Yousry A. El-Kassaby , Shu Diao , Qifu Luan , Yanjie Li

Advances in high-throughput phenotyping and genomics have accelerated our comprehension of plant functional differentiation. Nevertheless, efficiently phenotyping long-lived tree breeding populations and studying their dynamic response to field conditions remains a challenge, hindering genetic dissection and selective breeding efforts. This study refined and employed a newly developed high-efficiency unmanned aerial vehicle (UAV) imaging system to assess the temporal response of a slash pine (Pinus elliottii) breeding population in field conditions quantitatively over 2 years, identifying six strongly interrelated dynamic growth traits. In a genome-wide association study, 34 trait-associated loci explained between 1.1 % and –14.2 % of temporal phenotypic variation. These genes and regulatory loci influence signal reception, transduction, and transcriptional regulation networks in dynamic growth, impacting metabolic pathways such as cell membrane assembly, cell wall degradation, and cell differentiation. The enhanced UAV imaging system facilitates comprehensive analysis of dynamic growth response in trees, aiding in the discovery of informative alleles to unravel the genetic basis of complex phenotypic variation in conifers.

高通量表型和基因组学的进步加速了我们对植物功能分化的理解。然而，高效地分型长寿树木育种群体和研究它们对田间条件的动态响应仍然是一个挑战，这阻碍了遗传解剖和选择性育种的努力。本研究利用新开发的高效无人机（UAV）成像系统，对2年多的野外条件下湿地松（Pinus elliottii）繁殖群体的时间响应进行了定量评估，确定了6个密切相关的动态生长性状。在一项全基因组关联研究中，34个性状相关位点解释了1.1%至- 14.2%的时间表型变异。这些基因和调控位点影响动态生长中的信号接收、转导和转录调控网络，影响细胞膜组装、细胞壁降解和细胞分化等代谢途径。增强的无人机成像系统有助于全面分析树木的动态生长响应，有助于发现信息等位基因，揭示针叶树复杂表型变异的遗传基础。

{"title":"Leveraging close-range UAV phenotyping and GWAS for enhanced understanding of slash pine growth dynamics","authors":"Xianyin Ding , Pieter B. Pelser , Cong Xu , Ilga Porth , Mingming Cui , Yousry A. El-Kassaby , Shu Diao , Qifu Luan , Yanjie Li","doi":"10.1016/j.inpa.2025.07.002","DOIUrl":"10.1016/j.inpa.2025.07.002","url":null,"abstract":"<div><div>Advances in high-throughput phenotyping and genomics have accelerated our comprehension of plant functional differentiation. Nevertheless, efficiently phenotyping long-lived tree breeding populations and studying their dynamic response to field conditions remains a challenge, hindering genetic dissection and selective breeding efforts. This study refined and employed a newly developed high-efficiency unmanned aerial vehicle (UAV) imaging system to assess the temporal response of a slash pine (<em>Pinus elliottii</em>) breeding population in field conditions quantitatively over 2 years, identifying six strongly interrelated dynamic growth traits. In a genome-wide association study, 34 trait-associated loci explained between 1.1 % and –14.2 % of temporal phenotypic variation. These genes and regulatory loci influence signal reception, transduction, and transcriptional regulation networks in dynamic growth, impacting metabolic pathways such as cell membrane assembly, cell wall degradation, and cell differentiation. The enhanced UAV imaging system facilitates comprehensive analysis of dynamic growth response in trees, aiding in the discovery of informative alleles to unravel the genetic basis of complex phenotypic variation in conifers.</div></div>","PeriodicalId":53443,"journal":{"name":"Information Processing in Agriculture","volume":"12 4","pages":"Pages 550-564"},"PeriodicalIF":7.4,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145697904","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Winter wheat yield prediction using linear and nonlinear machine learning algorithms based on climatological and remote sensing data 基于气候和遥感数据的线性和非线性机器学习算法冬小麦产量预测

IF 7.4 Q1 AGRICULTURE, MULTIDISCIPLINARY

Information Processing in Agriculture

Pub Date : 2025-12-01 Epub Date: 2025-02-27 DOI: 10.1016/j.inpa.2025.02.004

Muhammad Haseeb , Zainab Tahir , Syed Amer Mahmood , Aqil Tariq

In the pursuit of enhancing agricultural forecasting in Pakistan, this research integrates remote sensing indices and climatic variables through advanced machine learning algorithms. By meticulously examining ten model combinations within different wheat season scenarios, the study employs nonlinear models, such as Random Forest (RF) and Support Vector Machines (SVM), and linear models, like Least Absolute Shrinkage and Selection Operator (LASSO) and Ridge. This research aims to predict wheat yield in Pakistan by integrating five remote sensing indices, including the Green Normalized Difference Vegetation Index (GNDVI), Normalized Difference Vegetation Index (NDVI), Enhanced Vegetation Index (EVI), Soil Adjusted Vegetation Index (SAVI), Atmospherically Resistant Vegetation Index (ARVI) with five climatic variables: Maximum Temperature (

T_{\max}

), Minimum Temperature (

T_{\min}

), Rainfall (R), Soil Moisture (SM), and Windspeed (WS) alongside the drought index and standardized Precipitation Evapotranspiration Index (SPEI). Ten model combinations were created within two wheat season scenarios: Full Seasonal Mean Scenario 1 (FSM) (SC1) and Peak Seasonal Mean Scenario 2 (PSM) (SC2). Two nonlinear ML algorithms, RF and SVM, and two linear models, LASSO and Ridge, were employed in both scenarios. Results indicated that in SC1, the RF model combination (GNDVI + SPEI + WS + SM) outperformed other models (R² = 0.75, RMSE = 2.40, MAE = 1.98). Similarly, in SC2, the RF regression surpassed SVM, with the model combination (GNDVI + SPEI + WS + SM) demonstrating the highest performance, achieving R² = 0.78, RMSE = 2.25, and MAE = 1.88, followed by (NDVI +

T_{\max}

+

T_{\min}

+ PPT + PET + WS + SM; R² = 0.75). The linear LASSO model also performed similarly to RF, achieving R² = 0.74–0.69 in both scenarios. The findings advocate for utilizing SC2 for yield prediction in ML models. Overall, this study underscores the significance and potential of ML methodologies in timely crop yield prediction across various crop growth stages, thereby establishing a robust foundation for ensuring regional food security.

为了提高巴基斯坦的农业预报能力，本研究通过先进的机器学习算法将遥感指数和气候变量相结合。通过仔细检查不同小麦季节情景下的10种模型组合，该研究采用非线性模型，如随机森林（RF）和支持向量机（SVM），以及线性模型，如最小绝对收缩和选择算子（LASSO）和Ridge。利用绿色归一化植被指数（GNDVI）、归一化植被指数（NDVI）、增强植被指数（EVI）、土壤调整植被指数（SAVI）、大气抗性植被指数（ARVI）等5个遥感指标，结合5个气候变量对巴基斯坦小麦产量进行预测。最高温度（Tmax）、最低温度（Tmin）、降雨量(R)、土壤湿度（SM）和风速（WS）以及干旱指数和标准化降水蒸散指数（SPEI）。在两个小麦季节情景中创建了10个模型组合：全季节平均情景1 (FSM) （SC1）和高峰季节平均情景2 (PSM) （SC2）。两种场景均采用了两种非线性ML算法RF和SVM，以及两种线性模型LASSO和Ridge。结果表明，在SC1中，RF模型组合（GNDVI + SPEI + WS + SM）优于其他模型（R2 = 0.75, RMSE = 2.40, MAE = 1.98）。同样，在SC2中，RF回归优于SVM，其中模型组合（GNDVI + SPEI + WS + SM）表现最好，R2 = 0.78, RMSE = 2.25, MAE = 1.88，其次是（NDVI + Tmax + Tmin + PPT + PET + WS + SM; R2 = 0.75）。线性LASSO模型的表现也与RF相似，在两种情况下均达到R2 = 0.74-0.69。研究结果提倡在ML模型中使用SC2进行产率预测。总体而言，本研究强调了ML方法在作物生长各个阶段及时预测作物产量方面的重要性和潜力，从而为确保区域粮食安全奠定了坚实的基础。

{"title":"Winter wheat yield prediction using linear and nonlinear machine learning algorithms based on climatological and remote sensing data","authors":"Muhammad Haseeb , Zainab Tahir , Syed Amer Mahmood , Aqil Tariq","doi":"10.1016/j.inpa.2025.02.004","DOIUrl":"10.1016/j.inpa.2025.02.004","url":null,"abstract":"<div><div>In the pursuit of enhancing agricultural forecasting in Pakistan, this research integrates remote sensing indices and climatic variables through advanced machine learning algorithms. By meticulously examining ten model combinations within different wheat season scenarios, the study employs nonlinear models, such as Random Forest (RF) and Support Vector Machines (SVM), and linear models, like Least Absolute Shrinkage and Selection Operator (LASSO) and Ridge. This research aims to predict wheat yield in Pakistan by integrating five remote sensing indices, including the Green Normalized Difference Vegetation Index (GNDVI), Normalized Difference Vegetation Index (NDVI), Enhanced Vegetation Index (EVI), Soil Adjusted Vegetation Index (SAVI), Atmospherically Resistant Vegetation Index (ARVI) with five climatic variables: Maximum Temperature (<span><math><msub><mi>T</mi><mrow><mi>max</mi></mrow></msub></math></span>), Minimum Temperature (<span><math><msub><mi>T</mi><mrow><mi>min</mi></mrow></msub></math></span>), Rainfall (R), Soil Moisture (SM), and Windspeed (WS) alongside the drought index and standardized Precipitation Evapotranspiration Index (SPEI). Ten model combinations were created within two wheat season scenarios: Full Seasonal Mean Scenario 1 (FSM) (SC1) and Peak Seasonal Mean Scenario 2 (PSM) (SC2). Two nonlinear ML algorithms, RF and SVM, and two linear models, LASSO and Ridge, were employed in both scenarios. Results indicated that in SC1, the RF model combination (GNDVI + SPEI + WS + SM) outperformed other models (R<sup>2</sup> = 0.75, RMSE = 2.40, MAE = 1.98). Similarly, in SC2, the RF regression surpassed SVM, with the model combination (GNDVI + SPEI + WS + SM) demonstrating the highest performance, achieving R<sup>2</sup> = 0.78, RMSE = 2.25, and MAE = 1.88, followed by (NDVI + <span><math><msub><mi>T</mi><mrow><mi>max</mi></mrow></msub></math></span> + <span><math><msub><mi>T</mi><mrow><mi>min</mi></mrow></msub></math></span> + PPT + PET + WS + SM; R<sup>2</sup> = 0.75). The linear LASSO model also performed similarly to RF, achieving R<sup>2</sup> = 0.74–0.69 in both scenarios. The findings advocate for utilizing SC2 for yield prediction in ML models. Overall, this study underscores the significance and potential of ML methodologies in timely crop yield prediction across various crop growth stages, thereby establishing a robust foundation for ensuring regional food security.</div></div>","PeriodicalId":53443,"journal":{"name":"Information Processing in Agriculture","volume":"12 4","pages":"Pages 431-444"},"PeriodicalIF":7.4,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145697950","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Automated detection of larval stages of the black soldier fly (Hermetia illucens Linnaeus) through deep learning augmented with optical flow 利用光流增强的深度学习技术自动检测黑兵蝇（Hermetia illucens Linnaeus）幼虫阶段

IF 7.4 Q1 AGRICULTURE, MULTIDISCIPLINARY

Information Processing in Agriculture

Pub Date : 2025-12-01 Epub Date: 2025-05-14 DOI: 10.1016/j.inpa.2025.05.001

Gianluca Manduca , Lloyd T. Wilson , Cesare Stefanini , Donato Romano

The black soldier fly (BSF) Hermetia illucens has garnered significant attention for its potential in sustainable waste management, nutrient recycling, and the production of valuable resources such as protein-rich animal feed and biofuels. Traditional mass production methods remain labor-intensive and error-prone, needing automated solutions. A critical challenge is the precise identification of BSF different life stages which is essential for optimizing feeding strategies, harvesting, and overall system efficiency. This study explores the use of deep learning, combined with optical flow analysis, to identify BSF life stages, particularly larvae, prepupae, and pupae. A Convolutional Neural Network (CNN) model was employed for real-time BSF larval stages detection. Training, validation, and test were performed on a comprehensive custom dataset of 2130 images. Evaluation metrics including precision, recall, and mean Average Precision (mAP) were assessed. Overall, the CNN model showed a precision of 0.96, a recall of 0.95, and a [email protected] of 0.97 on the test set, confirming its generalization capability and effectiveness in real-world scenarios. The integration of optical flow enhanced the model’s performance by leveraging prior knowledge of motor activity, particularly for identifying and correcting false positives in pupae classification. Automated identification of BSF larval stages optimizes resource management, reduces operational costs, and enhances the economic viability of BSF-based systems. The proposed system extends beyond terrestrial concerns, with potential implications for bioregenerative life-support systems, a promising space technology.

黑兵蝇（BSF） Hermetia illucens因其在可持续废物管理、养分回收和生产富含蛋白质的动物饲料和生物燃料等宝贵资源方面的潜力而引起了广泛关注。传统的批量生产方法仍然是劳动密集型的，容易出错，需要自动化的解决方案。一个关键的挑战是准确识别生物生物的不同生命阶段，这对于优化饲养策略、收获和整体系统效率至关重要。本研究探索了使用深度学习和光流分析相结合的方法来识别生物丝虫病的生命阶段，特别是幼虫、预蛹和蛹。采用卷积神经网络（CNN）模型实时检测BSF幼虫分期。训练、验证和测试是在2130张图像的综合定制数据集上进行的。评估指标包括精密度、召回率和平均平均精密度（mAP）。总体而言，CNN模型在测试集上的准确率为0.96，召回率为0.95，[email protected]为0.97，证实了其在现实场景中的泛化能力和有效性。光流的集成通过利用运动活动的先验知识增强了模型的性能，特别是在蛹分类中识别和纠正误报方面。BSF幼虫阶段的自动识别优化了资源管理，降低了运营成本，并提高了基于BSF的系统的经济可行性。拟议的系统超出了对地球的关注，对生物再生生命支持系统有潜在的影响，这是一项有前途的空间技术。

{"title":"Automated detection of larval stages of the black soldier fly (Hermetia illucens Linnaeus) through deep learning augmented with optical flow","authors":"Gianluca Manduca , Lloyd T. Wilson , Cesare Stefanini , Donato Romano","doi":"10.1016/j.inpa.2025.05.001","DOIUrl":"10.1016/j.inpa.2025.05.001","url":null,"abstract":"<div><div>The black soldier fly (BSF) <em>Hermetia illucens</em> has garnered significant attention for its potential in sustainable waste management, nutrient recycling, and the production of valuable resources such as protein-rich animal feed and biofuels. Traditional mass production methods remain labor-intensive and error-prone, needing automated solutions. A critical challenge is the precise identification of BSF different life stages which is essential for optimizing feeding strategies, harvesting, and overall system efficiency. This study explores the use of deep learning, combined with optical flow analysis, to identify BSF life stages, particularly larvae, prepupae, and pupae. A Convolutional Neural Network (CNN) model was employed for real-time BSF larval stages detection. Training, validation, and test were performed on a comprehensive custom dataset of 2130 images. Evaluation metrics including precision, recall, and mean Average Precision (mAP) were assessed. Overall, the CNN model showed a precision of 0.96, a recall of 0.95, and a [email protected] of 0.97 on the test set, confirming its generalization capability and effectiveness in real-world scenarios. The integration of optical flow enhanced the model’s performance by leveraging prior knowledge of motor activity, particularly for identifying and correcting false positives in pupae classification. Automated identification of BSF larval stages optimizes resource management, reduces operational costs, and enhances the economic viability of BSF-based systems. The proposed system extends beyond terrestrial concerns, with potential implications for bioregenerative life-support systems, a promising space technology.</div></div>","PeriodicalId":53443,"journal":{"name":"Information Processing in Agriculture","volume":"12 4","pages":"Pages 501-510"},"PeriodicalIF":7.4,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145697900","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Combining multiple spectral preprocessing and wavelength optimization methods improves potato aboveground biomass estimation 结合多光谱预处理和波长优化方法，改进了马铃薯地上生物量估算方法

IF 7.4 Q1 AGRICULTURE, MULTIDISCIPLINARY

Information Processing in Agriculture

Pub Date : 2025-12-01 Epub Date: 2025-06-26 DOI: 10.1016/j.inpa.2025.06.001

Yang Liu , Yiguang Fan , Jiejie Fan , Jibo Yue , Riqiang Chen , Yanpeng Ma , Mingbo Bian , Fuqin Yang , Haikuan Feng

Aboveground biomass (AGB) reflects the accumulation of crop photosynthesis, and AGB data guide agricultural production and field management practices. AGB can be estimated using UAV hyperspectral data; however, external factors and high-dimensional data lead to uncertainties. To address these issues, a cascading spectral preprocessing and band-optimized AGB estimation framework are proposed. We collected canopy hyperspectral reflectance and potato AGB data across two varieties, three planting densities, four nitrogen levels, and two potassium treatments during three growth stages. Then, we systematically compared the performance of Savitzky-Golay (SG) smoothing, multiplicative scatter correction (MSC), first-order differentiation (FOD) and their cascaded combinations. We also rigorously evaluated the ability of competitive adaptive reweighted sampling (CARS), successive projection algorithm (SPA) and their cascaded combination (CARS-SPA) to identify sensitive bands. The results indicated that cascaded spectral preprocessing methods significantly enhance the accuracy of potato AGB estimation. Among these approaches, the SG-MSC-FOD cascade performed most effectively. The combination of CARS and SPA yielded the fewest model variables while achieving the highest estimation accuracy. Furthermore, the integration of SG-MSC-FOD and CARS-SPA with partial least squares regression achieved the highest accuracy in AGB estimation across multiple growth stages, with a coefficient of determination (R²) of 0.73, root mean square error (RMSE) of 256.09 kg/hm², and normalized root mean square error (NRMSE) of 21.51 %. We validated the proposed method under different varieties, planting densities, and nitrogen and potassium treatments. This approach effectively reduces noise, lowers dimensionality, and enhances AGB estimation accuracy, providing a reliable solution for monitoring potato crop growth using hyperspectral remote sensing.

地上生物量（AGB）反映了作物光合作用的积累，AGB数据指导农业生产和田间管理实践。利用无人机高光谱数据估算AGB；然而，外部因素和高维数据导致了不确定性。为了解决这些问题，提出了一种级联光谱预处理和带优化AGB估计框架。本研究采集了2个品种、3种种植密度、4种氮素水平和2种钾肥处理在3个生育期的冠层高光谱反射率和马铃薯AGB数据。然后，系统地比较了Savitzky-Golay （SG）平滑、乘法散射校正（MSC）、一阶微分（FOD）及其级联组合的性能。我们还严格评估了竞争自适应重加权采样（CARS）、连续投影算法（SPA）及其级联组合（CARS-SPA）识别敏感波段的能力。结果表明，级联光谱预处理方法显著提高了马铃薯AGB估计的精度。在这些方法中，SG-MSC-FOD级联最有效。CARS和SPA的组合在获得最高估计精度的同时产生最少的模型变量。此外，SG-MSC-FOD和CARS-SPA结合偏最小二乘回归对多个生长阶段的AGB估计精度最高，决定系数（R2）为0.73，均方根误差（RMSE）为256.09 kg/hm2，归一化均方根误差（NRMSE）为21.51%。在不同品种、不同种植密度、不同氮钾处理条件下对该方法进行了验证。该方法有效地降低了噪声，降低了维数，提高了AGB估计精度，为马铃薯作物生长的高光谱遥感监测提供了可靠的解决方案。

{"title":"Combining multiple spectral preprocessing and wavelength optimization methods improves potato aboveground biomass estimation","authors":"Yang Liu , Yiguang Fan , Jiejie Fan , Jibo Yue , Riqiang Chen , Yanpeng Ma , Mingbo Bian , Fuqin Yang , Haikuan Feng","doi":"10.1016/j.inpa.2025.06.001","DOIUrl":"10.1016/j.inpa.2025.06.001","url":null,"abstract":"<div><div>Aboveground biomass (AGB) reflects the accumulation of crop photosynthesis, and AGB data guide agricultural production and field management practices. AGB can be estimated using UAV hyperspectral data; however, external factors and high-dimensional data lead to uncertainties. To address these issues, a cascading spectral preprocessing and band-optimized AGB estimation framework are proposed. We collected canopy hyperspectral reflectance and potato AGB data across two varieties, three planting densities, four nitrogen levels, and two potassium treatments during three growth stages. Then, we systematically compared the performance of Savitzky-Golay (SG) smoothing, multiplicative scatter correction (MSC), first-order differentiation (FOD) and their cascaded combinations. We also rigorously evaluated the ability of competitive adaptive reweighted sampling (CARS), successive projection algorithm (SPA) and their cascaded combination (CARS-SPA) to identify sensitive bands. The results indicated that cascaded spectral preprocessing methods significantly enhance the accuracy of potato AGB estimation. Among these approaches, the SG-MSC-FOD cascade performed most effectively. The combination of CARS and SPA yielded the fewest model variables while achieving the highest estimation accuracy. Furthermore, the integration of SG-MSC-FOD and CARS-SPA with partial least squares regression achieved the highest accuracy in AGB estimation across multiple growth stages, with a coefficient of determination (R<sup>2</sup>) of 0.73, root mean square error (RMSE) of 256.09 kg/hm<sup>2</sup>, and normalized root mean square error (NRMSE) of 21.51 %. We validated the proposed method under different varieties, planting densities, and nitrogen and potassium treatments. This approach effectively reduces noise, lowers dimensionality, and enhances AGB estimation accuracy, providing a reliable solution for monitoring potato crop growth using hyperspectral remote sensing.</div></div>","PeriodicalId":53443,"journal":{"name":"Information Processing in Agriculture","volume":"12 4","pages":"Pages 511-521"},"PeriodicalIF":7.4,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145697901","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

An integrated solution for collaborative scheduling of heterogeneous agricultural machines of different types in harvesting-transportation scenarios 采运场景下异构农机协同调度的集成解决方案

IF 7.4 Q1 AGRICULTURE, MULTIDISCIPLINARY

Information Processing in Agriculture

Pub Date : 2025-12-01 Epub Date: 2025-06-19 DOI: 10.1016/j.inpa.2025.06.002

Ning Wang , Zhiwen Jin , Man Zhang , Jianxing Xiao , Tianhai Wang , Qiang Sheng , Hao Wang , Han Li

Efficient coordination of machinery fleets in regional farmland operations remains a significant challenge due to the lack of scientifically grounded scheduling management strategies, high modeling complexity, and elevated operational costs. This study proposed an integrated solution for collaborative scheduling of heterogeneous agricultural machines of different types, aiming to address the collaborative scheduling of harvesters and grain trucks in harvest-transport scenarios. Firstly, an electronic farm map was constructed to facilitate path planning and generate unloading points within plots. The study then developed a collaborative scheduling model involving multiple machines, which incorporated heterogeneous parameters such as harvester harvesting speeds and grain truck hopper capacities. The model aims to minimize the total operational time of the machinery fleet. The scheduling problem was addressed by introducing a hybrid greedy heuristic-based improved genetic algorithm. Simulation and experimental validation were conducted using the electronic map of the Shanghai Qingpu unmanned farm. The results demonstrated that the proposed algorithm outperforms three algorithms in optimizing total operational time. For example, when the number of tasks is 20, the average total operational time is reduced by 32.4 min, an improvement of approximately 11.45% compared to the standard genetic algorithm. Additionally, parameter comparison experiments validate the algorithm’s compatibility with heterogeneous parameter settings, thereby substantiating its efficacy in addressing task allocation problems for heterogeneous machinery. The effectiveness of the proposed method in facilitating efficient collaboration among heterogeneous agricultural machines of different types is demonstrated through a case study on collaborative scheduling in harvest-transport scenarios. The findings validate the feasibility and applicability of the proposed approach in effectively addressing real-world agricultural scheduling challenges.

由于缺乏科学的调度管理策略，建模复杂性高，操作成本高，在区域农田作业中有效协调机队仍然是一个重大挑战。针对收割机与粮食运输车在收获运输场景下的协同调度问题，提出了一种不同类型异构农业机械协同调度的集成解决方案。首先，构建电子农场地图，方便路径规划和小区内卸载点的生成；然后，该研究开发了一个涉及多机器的协作调度模型，该模型包含了收割机收获速度和粮食卡车料斗容量等异构参数。该模型旨在使机队的总运行时间最小化。引入了一种基于混合贪心启发式的改进遗传算法来解决调度问题。利用上海青浦无人农场电子地图进行了仿真和实验验证。结果表明，该算法在优化总运行时间方面优于三种算法。例如，当任务数为20时，平均总操作时间减少了32.4 min，与标准遗传算法相比，提高了约11.45%。此外，参数对比实验验证了该算法对异构参数设置的兼容性，从而验证了该算法在解决异构机械任务分配问题方面的有效性。通过对收获-运输场景下协同调度的案例研究，证明了该方法在促进不同类型的异构农业机械之间高效协作方面的有效性。研究结果验证了该方法在有效解决现实世界农业调度挑战方面的可行性和适用性。

{"title":"An integrated solution for collaborative scheduling of heterogeneous agricultural machines of different types in harvesting-transportation scenarios","authors":"Ning Wang , Zhiwen Jin , Man Zhang , Jianxing Xiao , Tianhai Wang , Qiang Sheng , Hao Wang , Han Li","doi":"10.1016/j.inpa.2025.06.002","DOIUrl":"10.1016/j.inpa.2025.06.002","url":null,"abstract":"<div><div>Efficient coordination of machinery fleets in regional farmland operations remains a significant challenge due to the lack of scientifically grounded scheduling management strategies, high modeling complexity, and elevated operational costs. This study proposed an integrated solution for collaborative scheduling of heterogeneous agricultural machines of different types, aiming to address the collaborative scheduling of harvesters and grain trucks in harvest-transport scenarios. Firstly, an electronic farm map was constructed to facilitate path planning and generate unloading points within plots. The study then developed a collaborative scheduling model involving multiple machines, which incorporated heterogeneous parameters such as harvester harvesting speeds and grain truck hopper capacities. The model aims to minimize the total operational time of the machinery fleet. The scheduling problem was addressed by introducing a hybrid greedy heuristic-based improved genetic algorithm. Simulation and experimental validation were conducted using the electronic map of the Shanghai Qingpu unmanned farm. The results demonstrated that the proposed algorithm outperforms three algorithms in optimizing total operational time. For example, when the number of tasks is 20, the average total operational time is reduced by 32.4 min, an improvement of approximately 11.45% compared to the standard genetic algorithm. Additionally, parameter comparison experiments validate the algorithm’s compatibility with heterogeneous parameter settings, thereby substantiating its efficacy in addressing task allocation problems for heterogeneous machinery. The effectiveness of the proposed method in facilitating efficient collaboration among heterogeneous agricultural machines of different types is demonstrated through a case study on collaborative scheduling in harvest-transport scenarios. The findings validate the feasibility and applicability of the proposed approach in effectively addressing real-world agricultural scheduling challenges.</div></div>","PeriodicalId":53443,"journal":{"name":"Information Processing in Agriculture","volume":"12 4","pages":"Pages 522-538"},"PeriodicalIF":7.4,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145697902","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Efficient instance segmentation for strawberry in greenhouses using YOLOv8n-MCP on edge devices 在边缘设备上使用YOLOv8n-MCP对温室中的草莓进行有效的实例分割

IF 7.4 Q1 AGRICULTURE, MULTIDISCIPLINARY

Information Processing in Agriculture

Pub Date : 2025-12-01 Epub Date: 2025-07-16 DOI: 10.1016/j.inpa.2025.07.001

Xinhao Zhang , Guangpeng Zhang , Jiayi Wang , Jinqi Yang , Quanqu Ge , Ran Zhao , Yang Wang

The labor cost in agriculture is gradually increasing, making it necessary to develop robots for strawberry picking. These robots require accurate strawberry localization, which remains challenging using machine vision. While instance segmentation can improve positioning accuracy, current algorithms are inefficient on edge computing devices during robot navigation and ineffective for recognizing strawberries in elevated cultivation. This paper proposes an improved YOLOv8n model (YOLOv8n-MCP) optimized for edge computing during robot navigation. The network implements three key improvements: 1) MobileNetV3 as the backbone, enhancing strawberry feature extraction under varied lighting while reducing parameters and GFLOPs; 2) a new Cross-scale Feature Fusion Module (CCFM) as the Neck, improving detection of strawberries at varying distances; and 3) Partial Convolution (PConv) to enhance C2f and Head components, further reducing network parameters and GFLOPs while improving FPS. Experimental results show that compared to YOLOv8n, YOLOv8n-MCP reduces parameters by 69 %, GFLOPs by 56 %, and increases FPS by 42 %. Tests on Nvidia Jetson Xavier NX demonstrate that YOLOv8n-MCP achieves 49.5 FPS, significantly outperforming the original YOLOv8n’s 37.6 FPS, effectively meeting the requirements for strawberry instance segmentation during robot navigation with edge devices.

农业的人工成本正在逐渐增加，因此有必要开发草莓采摘机器人。这些机器人需要精确的草莓定位，这仍然是使用机器视觉的挑战。虽然实例分割可以提高定位精度，但目前的算法在机器人导航过程中的边缘计算设备上效率低下，对于识别高架栽培的草莓无效。本文提出了一种改进的YOLOv8n模型（YOLOv8n- mcp），对机器人导航过程中的边缘计算进行了优化。该网络实现了三个关键改进：1)以MobileNetV3为骨干，增强了不同光照下草莓特征的提取，同时降低了参数和GFLOPs；2)采用新的跨尺度特征融合模块（Cross-scale Feature Fusion Module， CCFM）作为颈部，提高草莓在不同距离上的检测能力；3)局部卷积（Partial Convolution, PConv）增强C2f和Head分量，进一步降低网络参数和GFLOPs，同时提高FPS。实验结果表明，与YOLOv8n相比，YOLOv8n- mcp可降低69%的参数，降低56%的GFLOPs，提高42%的FPS。在Nvidia Jetson Xavier NX上的测试表明，YOLOv8n- mcp达到49.5 FPS，显著优于原版YOLOv8n的37.6 FPS，有效满足机器人边缘设备导航时草莓实例分割的要求。

{"title":"Efficient instance segmentation for strawberry in greenhouses using YOLOv8n-MCP on edge devices","authors":"Xinhao Zhang , Guangpeng Zhang , Jiayi Wang , Jinqi Yang , Quanqu Ge , Ran Zhao , Yang Wang","doi":"10.1016/j.inpa.2025.07.001","DOIUrl":"10.1016/j.inpa.2025.07.001","url":null,"abstract":"<div><div>The labor cost in agriculture is gradually increasing, making it necessary to develop robots for strawberry picking. These robots require accurate strawberry localization, which remains challenging using machine vision. While instance segmentation can improve positioning accuracy, current algorithms are inefficient on edge computing devices during robot navigation and ineffective for recognizing strawberries in elevated cultivation. This paper proposes an improved YOLOv8n model (YOLOv8n-MCP) optimized for edge computing during robot navigation. The network implements three key improvements: 1) MobileNetV3 as the backbone, enhancing strawberry feature extraction under varied lighting while reducing parameters and GFLOPs; 2) a new Cross-scale Feature Fusion Module (CCFM) as the Neck, improving detection of strawberries at varying distances; and 3) Partial Convolution (PConv) to enhance C2f and Head components, further reducing network parameters and GFLOPs while improving FPS. Experimental results show that compared to YOLOv8n, YOLOv8n-MCP reduces parameters by 69 %, GFLOPs by 56 %, and increases FPS by 42 %. Tests on Nvidia Jetson Xavier NX demonstrate that YOLOv8n-MCP achieves 49.5 FPS, significantly outperforming the original YOLOv8n’s 37.6 FPS, effectively meeting the requirements for strawberry instance segmentation during robot navigation with edge devices.</div></div>","PeriodicalId":53443,"journal":{"name":"Information Processing in Agriculture","volume":"12 4","pages":"Pages 539-549"},"PeriodicalIF":7.4,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145697903","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Economics and barriers of precision viticulture technologies: A comprehensive systematic literature review 精准葡萄栽培技术的经济与障碍：全面系统的文献综述

IF 7.4 Q1 AGRICULTURE, MULTIDISCIPLINARY

Information Processing in Agriculture

Pub Date : 2025-12-01 Epub Date: 2025-04-11 DOI: 10.1016/j.inpa.2025.04.001

Antonino GALATI, Serena SOFIA, Maria CRESCIMANNO

Precision farming technologies are revolutionising the wine-growing sector thanks to their ability to manage crop variability, increase economic benefits, reduce the environmental impact, and improve grape yields and quality. Most earlier studies focused on the effects of precision technology adoption on plant health and canopy development—and therefore grape quality—neglecting the profitability impact. This study aims to fill this gap by presenting a systematic literature analysis discussing advancements in the economics of precision viticulture technologies. The results show how technologies such as unmanned aerial vehicles, precision irrigation, and robotics can increase efficiency in resource management, helping to reduce costs and improve vineyard profitability. However, the findings also emphasise the need for tailored approaches to integrate these advances. Furthermore, the analysis highlights the main barriers related to the cost of adopting precision technologies and the skills required to read and interpret the data. The results of this study hold interest to academics, vine growers, and farmers, providing a basis for future research into the cost-effectiveness of adopting precision technologies.

精准农业技术正在彻底改变葡萄酒种植行业，因为它们能够管理作物的变化，增加经济效益，减少对环境的影响，提高葡萄产量和质量。大多数早期的研究都集中在采用精密技术对植物健康和冠层发育的影响上，从而忽略了对盈利能力的影响。本研究旨在填补这一空白，提出了一个系统的文献分析，讨论经济的进步，精密葡萄栽培技术。研究结果表明，无人机、精准灌溉和机器人技术等技术可以提高资源管理效率，帮助降低成本，提高葡萄园的盈利能力。然而，研究结果也强调需要有针对性的方法来整合这些进步。此外，分析强调了与采用精密技术的成本和阅读和解释数据所需的技能有关的主要障碍。这项研究的结果引起了学术界、葡萄种植者和农民的兴趣，为未来研究采用精密技术的成本效益提供了基础。

{"title":"Economics and barriers of precision viticulture technologies: A comprehensive systematic literature review","authors":"Antonino GALATI, Serena SOFIA, Maria CRESCIMANNO","doi":"10.1016/j.inpa.2025.04.001","DOIUrl":"10.1016/j.inpa.2025.04.001","url":null,"abstract":"<div><div>Precision farming technologies are revolutionising the wine-growing sector thanks to their ability to manage crop variability, increase economic benefits, reduce the environmental impact, and improve grape yields and quality. Most earlier studies focused on the effects of precision technology adoption on plant health and canopy development—and therefore grape quality—neglecting the profitability impact. This study aims to fill this gap by presenting a systematic literature analysis discussing advancements in the economics of precision viticulture technologies. The results show how technologies such as unmanned aerial vehicles, precision irrigation, and robotics can increase efficiency in resource management, helping to reduce costs and improve vineyard profitability. However, the findings also emphasise the need for tailored approaches to integrate these advances. Furthermore, the analysis highlights the main barriers related to the cost of adopting precision technologies and the skills required to read and interpret the data. The results of this study hold interest to academics, vine growers, and farmers, providing a basis for future research into the cost-effectiveness of adopting precision technologies.</div></div>","PeriodicalId":53443,"journal":{"name":"Information Processing in Agriculture","volume":"12 4","pages":"Pages 487-500"},"PeriodicalIF":7.4,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145697899","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Detection of fungal disease in citrus fruit based on hyperspectral imaging 基于高光谱成像的柑橘果实真菌病害检测

IF 7.4 Q1 AGRICULTURE, MULTIDISCIPLINARY

Information Processing in Agriculture

Pub Date : 2025-12-01 Epub Date: 2025-03-01 DOI: 10.1016/j.inpa.2025.02.006

Xincai Yu , Shuangyin Liu , Chenjiaozi Wang , Binbin Jiao , Cong Huang , Bo Liu , Conghui Liu , Liping Yin , Fanghao Wan , Wanqiang Qian , Xi Qiao

Citrus fruit fungal disease is a major reason for the serious decline in citrus production and quality. Due to its highly contagious nature, timely and effective detection is an important means of prevention and control. Given the high similarity between citrus quarantine diseases and local similar diseases after invading citrus fruits, this study utilizes hyperspectral imaging technology to acquire hyperspectral images of citrus diseases caused by three types of fungi (Phytophthora citrophthora, Phytophthora citricola, Phytophthora syringae). By studying the spectral features of different regions affected by citrus diseases, the competitive adaptive resampling algorithm (CARS) was used to extract 44 feature bands for reconstructing the spectral image, aiming to reduce information redundancy without losing critical information. A simple deep learning model architecture was proposed, which achieved an accuracy of 92.50% in the test dataset. This study provides a new perspective and method for citrus disease detection, offering theoretical and scientific support for the detection of citrus diseases using deep learning and hyperspectral imaging technology.

柑桔果实真菌病是柑桔产量和品质严重下降的主要原因。由于其高传染性，及时有效的检测是预防和控制的重要手段。鉴于柑橘检疫性疾病与柑橘果实侵染后的当地类似疾病高度相似，本研究利用高光谱成像技术获取了柑橘疫霉（Phytophthora citrophthora）、citricola疫霉（Phytophthora citricola）、丁香疫霉（Phytophthora syringae）三种真菌引起的柑橘病害的高光谱图像。通过研究柑橘病害不同区域的光谱特征，采用竞争自适应重采样算法（CARS）提取44个特征波段进行光谱图像重构，在不丢失关键信息的前提下减少信息冗余。提出了一种简单的深度学习模型架构，在测试数据集中达到了92.50%的准确率。本研究为柑橘病害检测提供了新的视角和方法，为利用深度学习和高光谱成像技术检测柑橘病害提供了理论和科学支持。

{"title":"Detection of fungal disease in citrus fruit based on hyperspectral imaging","authors":"Xincai Yu , Shuangyin Liu , Chenjiaozi Wang , Binbin Jiao , Cong Huang , Bo Liu , Conghui Liu , Liping Yin , Fanghao Wan , Wanqiang Qian , Xi Qiao","doi":"10.1016/j.inpa.2025.02.006","DOIUrl":"10.1016/j.inpa.2025.02.006","url":null,"abstract":"<div><div>Citrus fruit fungal disease is a major reason for the serious decline in citrus production and quality. Due to its highly contagious nature, timely and effective detection is an important means of prevention and control. Given the high similarity between citrus quarantine diseases and local similar diseases after invading citrus fruits, this study utilizes hyperspectral imaging technology to acquire hyperspectral images of citrus diseases caused by three types of fungi (<em>Phytophthora citrophthora</em>, <em>Phytophthora citricola</em>, <em>Phytophthora syringae</em>). By studying the spectral features of different regions affected by citrus diseases, the competitive adaptive resampling algorithm (CARS) was used to extract 44 feature bands for reconstructing the spectral image, aiming to reduce information redundancy without losing critical information. A simple deep learning model architecture was proposed, which achieved an accuracy of 92.50% in the test dataset. This study provides a new perspective and method for citrus disease detection, offering theoretical and scientific support for the detection of citrus diseases using deep learning and hyperspectral imaging technology.</div></div>","PeriodicalId":53443,"journal":{"name":"Information Processing in Agriculture","volume":"12 4","pages":"Pages 456-465"},"PeriodicalIF":7.4,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145697843","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Lightweight precision model for watermelon seed group density estimation and counting 西瓜种子群密度估算与计数的轻量级精度模型

IF 7.4 Q1 AGRICULTURE, MULTIDISCIPLINARY

Information Processing in Agriculture

Pub Date : 2025-12-01 Epub Date: 2025-07-24 DOI: 10.1016/j.inpa.2025.07.003

Helong Yu , Liyun Han , Chengcheng Chen , Honghong Su , Qichao Niu , Ronghao Meng , Mingxuan Xue

The accurate counting of overlapping watermelon seeds is a key foundation for seed quality testing, breeding selection, resource allocation, and other processes. To improve the counting accuracy for flat and slightly overlapping seeds, we introduce LOYOLO-GC, a Lightweight Occlusion YOLO8n-based group counting model. It adopts HGNetV2 as its backbone, where HGBlocks extract multi-level features for improved learning. GhostConv replaces the standard convolution in HGBlocks, forming LightHGBlock to reduce the number of parameters by generating intrinsic and ghost feature maps with fewer kernels. In addition, a Large Separable Kernel Attention mechanism (LSKA) is used to decompose deep convolution kernels into horizontal and vertical 1D kernels, enabling efficient large kernel attention with lower computational and memory cost. After optimizing the model, we build a multi-occlusion watermelon seed dataset and employ it to develop a LOYOLO-based group counting method. The experimental results show that LOYOLO-GC outperforms SOTA models, achieving 96.08 % accuracy and 86.66 % mAP, an improvement of 0.48 % and 1.67 %, respectively. The model parameters decrease by 63.8 % and GMACs decrease by 38.9 %. Counting accuracy is also improved, with ACC increasing by 5.32 % and L-ACC increasing by 5.04 %, while MAE and RMSE are decreased by 3.68 and 3.28, respectively.

西瓜重叠种子的准确计数是西瓜种子质量检测、育种选择、资源配置等过程的重要基础。为了提高平坦和轻微重叠种子的计数精度，我们引入了基于轻量级遮挡yolo8n的分组计数模型LOYOLO-GC。它采用HGNetV2作为主干，其中HGBlocks提取多层次特征以提高学习。GhostConv取代HGBlocks中的标准卷积，形成LightHGBlock，通过生成内核更少的内在和幽灵特征映射来减少参数的数量。此外，采用大可分离核注意机制（Large分离式核注意机制，LSKA）将深度卷积核分解为水平和垂直的一维核，在降低计算和内存成本的同时实现高效的大核注意。在对模型进行优化后，我们构建了一个多遮挡的西瓜种子数据集，并利用该数据集开发了一种基于loyolo的分组计数方法。实验结果表明，LOYOLO-GC模型的准确率为96.08%，mAP的准确率为86.66%，分别比SOTA模型提高0.48%和1.67%。模型参数降低63.8%，gmac降低38.9%。计数精度也有所提高，ACC提高了5.32%，L-ACC提高了5.04%，MAE和RMSE分别降低了3.68和3.28。

{"title":"Lightweight precision model for watermelon seed group density estimation and counting","authors":"Helong Yu , Liyun Han , Chengcheng Chen , Honghong Su , Qichao Niu , Ronghao Meng , Mingxuan Xue","doi":"10.1016/j.inpa.2025.07.003","DOIUrl":"10.1016/j.inpa.2025.07.003","url":null,"abstract":"<div><div>The accurate counting of overlapping watermelon seeds is a key foundation for seed quality testing, breeding selection, resource allocation, and other processes. To improve the counting accuracy for flat and slightly overlapping seeds, we introduce LOYOLO-GC, a Lightweight Occlusion YOLO8n-based group counting model. It adopts HGNetV2 as its backbone, where HGBlocks extract multi-level features for improved learning. GhostConv replaces the standard convolution in HGBlocks, forming LightHGBlock to reduce the number of parameters by generating intrinsic and ghost feature maps with fewer kernels. In addition, a Large Separable Kernel Attention mechanism (LSKA) is used to decompose deep convolution kernels into horizontal and vertical 1D kernels, enabling efficient large kernel attention with lower computational and memory cost. After optimizing the model, we build a multi-occlusion watermelon seed dataset and employ it to develop a LOYOLO-based group counting method. The experimental results show that LOYOLO-GC outperforms SOTA models, achieving 96.08 % accuracy and 86.66 % mAP, an improvement of 0.48 % and 1.67 %, respectively. The model parameters decrease by 63.8 % and GMACs decrease by 38.9 %. Counting accuracy is also improved, with ACC increasing by 5.32 % and L-ACC increasing by 5.04 %, while MAE and RMSE are decreased by 3.68 and 3.28, respectively.</div></div>","PeriodicalId":53443,"journal":{"name":"Information Processing in Agriculture","volume":"12 4","pages":"Pages 565-580"},"PeriodicalIF":7.4,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145697905","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0