Information Processing in Agriculture最新文献_第3页

Enhancing leaf disease detection accuracy through synergistic integration of deep transfer learning and multimodal techniques 通过深度迁移学习和多模态技术的协同集成，提高叶片病害检测的准确性

IF 7.4 Q1 AGRICULTURE, MULTIDISCIPLINARY

Information Processing in Agriculture

Pub Date : 2025-09-01 DOI: 10.1016/j.inpa.2024.09.006

Divine Senanu Ametefe , Suzi Seroja Sarnin , Darmawaty Mohd Ali , Aziz Caliskan , Imène Tatar Caliskan , Abdulmalik Adozuka Aliu , Dah John

The agricultural sector, a cornerstone of economies worldwide, faces significant challenges due to plant diseases, which severely affect crop yield and quality. Early and accurate detection of these diseases is crucial for effective mitigation strategies. The current methods used often lack accuracy and adaptability, especially in diverse environmental conditions. This study introduces a novel, synergistic approach that integrates deep transfer learning with multimodal techniques, specifically canny edges, colour spectrum intensity analysis, and custom data augmentation strategies. Unlike existing methods that rely solely on pre-trained models, the approach utilised in this study offers an innovative fusion of distinct feature extraction techniques. The canny edges highlighted the structural intricacies of leaf diseases, while colour spectrum intensity analysis enhanced the detection of disease-specific colour markers. The customized data augmentation techniques employed (in the study) was shown to enhance the learning process of the models, resulting in their adaptability to diverse agricultural environments. This integration applied to DenseNet201 and EfficientNetB3, achieved detection accuracies of 99.03 % and 98.23 %, respectively, surpassing traditional models and setting new benchmarks in plant disease detection. These results demonstrate the effectiveness of the proposed multi-faceted approach and its potential to significantly enhance crop disease management systems.

农业部门是全球经济的基石，但由于植物病害而面临重大挑战，严重影响作物产量和质量。早期和准确发现这些疾病对于有效的缓解战略至关重要。目前使用的方法往往缺乏准确性和适应性，特别是在不同的环境条件下。本研究介绍了一种新颖的协同方法，该方法将深度迁移学习与多模态技术相结合，特别是精细边缘、光谱强度分析和自定义数据增强策略。与仅依赖预训练模型的现有方法不同，本研究中使用的方法提供了不同特征提取技术的创新融合。巧妙的边缘突出了叶片疾病的结构复杂性，而光谱强度分析增强了对疾病特异性颜色标记的检测。研究中采用的定制数据增强技术可以增强模型的学习过程，从而使模型能够适应不同的农业环境。将该集成应用于DenseNet201和EfficientNetB3，检测准确率分别达到99.03%和98.23%，超越了传统模型，为植物病害检测树立了新的基准。这些结果证明了所提出的多方面方法的有效性及其显著增强作物病害管理系统的潜力。

{"title":"Enhancing leaf disease detection accuracy through synergistic integration of deep transfer learning and multimodal techniques","authors":"Divine Senanu Ametefe , Suzi Seroja Sarnin , Darmawaty Mohd Ali , Aziz Caliskan , Imène Tatar Caliskan , Abdulmalik Adozuka Aliu , Dah John","doi":"10.1016/j.inpa.2024.09.006","DOIUrl":"10.1016/j.inpa.2024.09.006","url":null,"abstract":"<div><div>The agricultural sector, a cornerstone of economies worldwide, faces significant challenges due to plant diseases, which severely affect crop yield and quality. Early and accurate detection of these diseases is crucial for effective mitigation strategies. The current methods used often lack accuracy and adaptability, especially in diverse environmental conditions. This study introduces a novel, synergistic approach that integrates deep transfer learning with multimodal techniques, specifically canny edges, colour spectrum intensity analysis, and custom data augmentation strategies. Unlike existing methods that rely solely on pre-trained models, the approach utilised in this study offers an innovative fusion of distinct feature extraction techniques. The canny edges highlighted the structural intricacies of leaf diseases, while colour spectrum intensity analysis enhanced the detection of disease-specific colour markers. The customized data augmentation techniques employed (in the study) was shown to enhance the learning process of the models, resulting in their adaptability to diverse agricultural environments. This integration applied to DenseNet201 and EfficientNetB3, achieved detection accuracies of 99.03 % and 98.23 %, respectively, surpassing traditional models and setting new benchmarks in plant disease detection. These results demonstrate the effectiveness of the proposed multi-faceted approach and its potential to significantly enhance crop disease management systems.</div></div>","PeriodicalId":53443,"journal":{"name":"Information Processing in Agriculture","volume":"12 3","pages":"Pages 279-299"},"PeriodicalIF":7.4,"publicationDate":"2025-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145327146","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

A novel method to detect stem and fruit dynamically for apricot posture estimation and adjustment 一种用于杏子姿态估计与调整的动态检测方法

IF 7.4 Q1 AGRICULTURE, MULTIDISCIPLINARY

Information Processing in Agriculture

Pub Date : 2025-09-01 DOI: 10.1016/j.inpa.2024.12.001

Wulan Mao , Leilei He , Man Xia , Hanhui Jiang , Rui Li , Ramesh Sahni , Yaqoob Majeed , Zhanjiang Zhu , Longsheng Fu

Separating pulp and core is critical for apricot processing, but faces labor shortages. To address this challenge, a fully automated pitting machine (FAPM) based on automatic apricot orientation device (AAOD) was proposed to achieve mechanized pitting by apricot automatic orientation. The designed and constructed AAOD adopt with dynamic visual detection and mechanical orientation for apricot posture adjustment. YOLOv8 series models were applied for apricot and stem detection, and then estimating their three-dimensional posture. Compared with other YOLOv8 series models, YOLOv8n was selected as the preferred detection model with a detection speed of 10.3 ms and a size of 6.1 MB to meet the need of real-time detection and lightweight deployment. YOLOv8n achieved precision (P), recall (R), and mean average precision (mAP) values of 82.0 %, 90.9 %, and 90.1 %, respectively. Moreover, new indicators, namely positional offsets in the image coordinate system (Offset_img), positional offsets (Offset_3D), angular offsets in the 3D coordinate system (Offset_ang), and the ratio of intersection to manual bounding box areas (Ratio_im), were proposed to validate the performance of AAOD for position estimation in three varieties of apricot. The best performance was obtained in Saimaiti apricot and achieved Offset_img of 2.9 pixels, Offset_3D of 1.2 mm, and Offset_ang of 0.9°, with Ratio_im for apricot and stem were 99.3 % and 97.3 %. Experimental show that the optimal operating parameters for AAOD are 20 rps for alignment wheel rotation speed and the distance of 22.5 mm from apricot base to alignment wheel axis, which presented the best successful orientation rate of 91.5 % with an Offset_3D of 1.8 mm. Result demonstrated that the dynamic detection-based orientation approach proposed in this study has great potential for automatic apricot pitting.

分离果肉和核对杏的加工至关重要，但面临劳动力短缺。为了解决这一问题，提出了一种基于杏自动定向装置（AAOD）的全自动点蚀机（FAPM），通过杏自动定向实现机械化点蚀。设计构建的AAOD采用动态视觉检测和机械定向的方式进行杏子姿态调整。应用YOLOv8系列模型对杏和茎进行检测，并对其三维姿态进行估计。与其他YOLOv8系列型号相比，为了满足实时检测和轻量化部署的需要，我们选择了YOLOv8n作为首选检测型号，检测速度为10.3 ms，大小为6.1 MB。YOLOv8n的精密度(P)、召回率(R)和平均精密度（mAP）分别为82.0%、90.9%和90.1%。此外，提出了新的指标，即图像坐标系中的位置偏移量（Offsetimg）、位置偏移量（Offset3D）、三维坐标系中的角度偏移量（Offsetang）以及相交与人工边界框面积的比值（Ratioim），验证了AAOD在杏的位置估计中的性能。以赛迈提杏为例，实现了2.9像素的偏移，偏移量为1.2 mm，偏移量为0.9°，其中杏和茎的偏移率分别为99.3%和97.3%。实验结果表明，AAOD的最佳工作参数为定位轮转速为20 rpm，杏座距定位轮轴线距离为22.5 mm，当偏移量为1.8 mm时，定位成功率为91.5%。结果表明，本文提出的基于动态检测的定向方法在杏点蚀自动检测中具有很大的应用潜力。

{"title":"A novel method to detect stem and fruit dynamically for apricot posture estimation and adjustment","authors":"Wulan Mao , Leilei He , Man Xia , Hanhui Jiang , Rui Li , Ramesh Sahni , Yaqoob Majeed , Zhanjiang Zhu , Longsheng Fu","doi":"10.1016/j.inpa.2024.12.001","DOIUrl":"10.1016/j.inpa.2024.12.001","url":null,"abstract":"<div><div>Separating pulp and core is critical for apricot processing, but faces labor shortages. To address this challenge, a fully automated pitting machine (FAPM) based on automatic apricot orientation device (AAOD) was proposed to achieve mechanized pitting by apricot automatic orientation. The designed and constructed AAOD adopt with dynamic visual detection and mechanical orientation for apricot posture adjustment. YOLOv8 series models were applied for apricot and stem detection, and then estimating their three-dimensional posture. Compared with other YOLOv8 series models, YOLOv8n was selected as the preferred detection model with a detection speed of 10.3 ms and a size of 6.1 MB to meet the need of real-time detection and lightweight deployment. YOLOv8n achieved precision (<em>P</em>), recall (<em>R</em>), and mean average precision (<em>mAP</em>) values of 82.0 %, 90.9 %, and 90.1 %, respectively. Moreover, new indicators, namely positional offsets in the image coordinate system (<em>Offset<sub>img</sub></em>), positional offsets (<em>Offset<sub>3D</sub></em>), angular offsets in the 3D coordinate system (<em>Offset<sub>ang</sub></em>), and the ratio of intersection to manual bounding box areas (<em>Ratio<sub>im</sub></em>), were proposed to validate the performance of AAOD for position estimation in three varieties of apricot. The best performance was obtained in Saimaiti apricot and achieved <em>Offset<sub>img</sub></em> of 2.9 pixels, <em>Offset<sub>3D</sub></em> of 1.2 mm, and <em>Offset<sub>ang</sub></em> of 0.9°, with <em>Ratio<sub>im</sub></em> for apricot and stem were 99.3 % and 97.3 %. Experimental show that the optimal operating parameters for AAOD are 20 rps for alignment wheel rotation speed and the distance of 22.5 mm from apricot base to alignment wheel axis, which presented the best successful orientation rate of 91.5 % with an <em>Offset<sub>3D</sub></em> of 1.8 mm. Result demonstrated that the dynamic detection-based orientation approach proposed in this study has great potential for automatic apricot pitting.</div></div>","PeriodicalId":53443,"journal":{"name":"Information Processing in Agriculture","volume":"12 3","pages":"Pages 358-369"},"PeriodicalIF":7.4,"publicationDate":"2025-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145327150","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

Integrating Artificial Intelligence in dairy farm management − biometric facial recognition for cows 将人工智能整合到奶牛场管理中——奶牛的生物面部识别

IF 7.4 Q1 AGRICULTURE, MULTIDISCIPLINARY

Information Processing in Agriculture

Pub Date : 2025-09-01 DOI: 10.1016/j.inpa.2024.10.001

Shubhangi Mahato , Suresh Neethirajan

The integration of Artificial Intelligence (AI) into dairy farm management through biometric facial recognition of cows marks a significant milestone in livestock care. This comprehensive review explores the development, implementation, and challenges associated with AI-powered biometric facial identification in dairy agriculture. It emphasizes the pivotal role of this innovation in enabling precise monitoring of individual cows, thereby facilitating thorough tracking of their health, behaviors, and productivity levels. Derived from facial recognition technologies originally designed for humans, this approach harnesses distinctive features of cow faces for gentle and immediate observation within large-scale farming operations. The evolution of AI from basic pattern recognition to advanced Convolutional Neural Networks (CNNs) and deep learning frameworks signifies a transition toward data-driven agriculture. This analysis addresses notable challenges such as environmental variability, data collection difficulties, ethical considerations, and technological limitations. Furthermore, it compares various AI frameworks, highlighting their unique advantages and suitability in the dairy farming context. Despite these obstacles, facial recognition technology holds promise for enhancing farm efficiency, improving animal welfare, and promoting sustainable practices, underscoring the need for ongoing research and innovation. We advocate for future investigations focused on enhancing adaptability to diverse environments, ensuring ethical AI deployment, fostering compatibility across different breeds, and integrating with complementary agricultural technologies. Ultimately, this review underscores the transformative impact of AI in advancing dairy farming towards a data-centric future while prioritizing responsible agricultural practices.

通过对奶牛进行生物面部识别，将人工智能（AI）整合到奶牛场管理中，这是牲畜护理领域的一个重要里程碑。这篇全面的综述探讨了与奶牛农业中人工智能生物识别面部识别相关的发展、实施和挑战。它强调了这项创新在实现对单个奶牛的精确监测方面的关键作用，从而促进了对它们的健康、行为和生产力水平的彻底跟踪。这种方法源自最初为人类设计的面部识别技术，利用奶牛面部的独特特征，在大规模农业经营中进行温和而迅速的观察。人工智能从基本模式识别到高级卷积神经网络（cnn）和深度学习框架的演变标志着向数据驱动农业的过渡。该分析解决了一些值得注意的挑战，如环境可变性、数据收集困难、伦理考虑和技术限制。此外，它比较了各种人工智能框架，突出了它们在奶牛养殖环境中的独特优势和适用性。尽管存在这些障碍，面部识别技术仍有望提高农场效率，改善动物福利，促进可持续实践，强调了持续研究和创新的必要性。我们主张未来的研究重点是提高对不同环境的适应性，确保人工智能的道德部署，促进不同品种的兼容性，以及与互补的农业技术相结合。最后，本综述强调了人工智能在推动奶业向以数据为中心的未来发展方面的变革性影响，同时优先考虑负责任的农业实践。

{"title":"Integrating Artificial Intelligence in dairy farm management − biometric facial recognition for cows","authors":"Shubhangi Mahato , Suresh Neethirajan","doi":"10.1016/j.inpa.2024.10.001","DOIUrl":"10.1016/j.inpa.2024.10.001","url":null,"abstract":"<div><div>The integration of Artificial Intelligence (AI) into dairy farm management through biometric facial recognition of cows marks a significant milestone in livestock care. This comprehensive review explores the development, implementation, and challenges associated with AI-powered biometric facial identification in dairy agriculture. It emphasizes the pivotal role of this innovation in enabling precise monitoring of individual cows, thereby facilitating thorough tracking of their health, behaviors, and productivity levels. Derived from facial recognition technologies originally designed for humans, this approach harnesses distinctive features of cow faces for gentle and immediate observation within large-scale farming operations. The evolution of AI from basic pattern recognition to advanced Convolutional Neural Networks (CNNs) and deep learning frameworks signifies a transition toward data-driven agriculture. This analysis addresses notable challenges such as environmental variability, data collection difficulties, ethical considerations, and technological limitations. Furthermore, it compares various AI frameworks, highlighting their unique advantages and suitability in the dairy farming context. Despite these obstacles, facial recognition technology holds promise for enhancing farm efficiency, improving animal welfare, and promoting sustainable practices, underscoring the need for ongoing research and innovation. We advocate for future investigations focused on enhancing adaptability to diverse environments, ensuring ethical AI deployment, fostering compatibility across different breeds, and integrating with complementary agricultural technologies. Ultimately, this review underscores the transformative impact of AI in advancing dairy farming towards a data-centric future while prioritizing responsible agricultural practices.</div></div>","PeriodicalId":53443,"journal":{"name":"Information Processing in Agriculture","volume":"12 3","pages":"Pages 312-325"},"PeriodicalIF":7.4,"publicationDate":"2025-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145327415","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 data from plant monitoring into crop models 利用来自植物监测的数据到作物模型中

IF 7.4 Q1 AGRICULTURE, MULTIDISCIPLINARY

Information Processing in Agriculture

Pub Date : 2025-09-01 DOI: 10.1016/j.inpa.2025.02.003

Monique Pires Gravina de Oliveira , Thais Queiroz Zorzeto-Cesar , Romis Ribeiro de Faissol Attux , Luiz Henrique Antunes Rodrigues

An increase in data availability from different sensors and sources has changed how crop models are being used. Data assimilation is one approach for integrating data and models that has been widely used for field crops but not yet in protected environments. We present a case study of data assimilation in a greenhouse, updating growth estimates of the Reduced State TOMGRO model. We assimilated data obtained through the continuous monitoring of plant mass and images captured by low-cost cameras, using the Unscented Kalman Filter and the Ensemble Kalman Filter. In some cases, assimilation led to improvements of more than 40% in the RMSE of yield estimates of the non-calibrated model, within a validation set. The improvements were more noticeable when there was a need to adjust the estimates to a condition the model does not represent. In these situations, we noted the RMSE decreased by almost 80%, depending on the variable being assimilated. However, in some cases, the results were also impaired by assimilation, and we highlight the impacts on the filter performance caused by the quality of observations and of observation models. Overall, the employed measurements, i.e., area of organs observed in pictures and plant-water mass, seemed suitable for tracking plant growth and for obtaining good approximations of the state variables estimated by the model. As with other studies, it was not the case that assimilating one state was useful for improving the value of others, including yield. As the first study using filters and non-destructive observations in a process-based crop model in a protected environment, we identified a lot of potential, but to identify the best use of these techniques with real-time data, more studies are needed. By making all data and code from this study available, we hope to ease future research in this area.

来自不同传感器和来源的可用数据的增加改变了作物模型的使用方式。数据同化是一种集成数据和模型的方法，已广泛应用于大田作物，但尚未应用于受保护环境。我们提出了一个温室数据同化的案例研究，更新了减少状态TOMGRO模型的增长估计。我们利用Unscented卡尔曼滤波和集合卡尔曼滤波对通过连续监测植物质量获得的数据和低成本相机拍摄的图像进行了同化。在某些情况下，在验证集中，同化导致非校准模型产量估计的RMSE提高40%以上。当需要调整估计以适应模型不代表的情况时，改进更加明显。在这些情况下，我们注意到RMSE下降了近80%，这取决于被吸收的变量。然而，在某些情况下，结果也会受到同化的影响，我们强调了观测质量和观测模型对滤波性能的影响。总的来说，所采用的测量，即在图片中观察到的器官面积和植物水分质量，似乎适合于跟踪植物生长，并获得由模型估计的状态变量的良好近似值。与其他研究一样，同化一种状态并不有助于提高其他状态的价值，包括产量。作为第一个在受保护的环境中使用过滤器和非破坏性观测的基于过程的作物模型的研究，我们发现了很多潜力，但为了确定这些技术与实时数据的最佳使用，还需要更多的研究。通过提供本研究的所有数据和代码，我们希望能够简化该领域的未来研究。

{"title":"Leveraging data from plant monitoring into crop models","authors":"Monique Pires Gravina de Oliveira , Thais Queiroz Zorzeto-Cesar , Romis Ribeiro de Faissol Attux , Luiz Henrique Antunes Rodrigues","doi":"10.1016/j.inpa.2025.02.003","DOIUrl":"10.1016/j.inpa.2025.02.003","url":null,"abstract":"<div><div>An increase in data availability from different sensors and sources has changed how crop models are being used. Data assimilation is one approach for integrating data and models that has been widely used for field crops but not yet in protected environments. We present a case study of data assimilation in a greenhouse, updating growth estimates of the Reduced State TOMGRO model. We assimilated data obtained through the continuous monitoring of plant mass and images captured by low-cost cameras, using the Unscented Kalman Filter and the Ensemble Kalman Filter. In some cases, assimilation led to improvements of more than 40% in the RMSE of yield estimates of the non-calibrated model, within a validation set. The improvements were more noticeable when there was a need to adjust the estimates to a condition the model does not represent. In these situations, we noted the RMSE decreased by almost 80%, depending on the variable being assimilated. However, in some cases, the results were also impaired by assimilation, and we highlight the impacts on the filter performance caused by the quality of observations and of observation models. Overall, the employed measurements, i.e., area of organs observed in pictures and plant-water mass, seemed suitable for tracking plant growth and for obtaining good approximations of the state variables estimated by the model. As with other studies, it was not the case that assimilating one state was useful for improving the value of others, including yield. As the first study using filters and non-destructive observations in a process-based crop model in a protected environment, we identified a lot of potential, but to identify the best use of these techniques with real-time data, more studies are needed. By making all data and code from this study available, we hope to ease future research in this area.</div></div>","PeriodicalId":53443,"journal":{"name":"Information Processing in Agriculture","volume":"12 3","pages":"Pages 408-429"},"PeriodicalIF":7.4,"publicationDate":"2025-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145327148","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 automatic method for estimating insect defoliation with visual highlights of consumed leaf tissue regions 一种利用消耗叶片组织区域的视觉亮点估算昆虫落叶情况的自动方法

IF 7.7 Q1 AGRICULTURE, MULTIDISCIPLINARY

Information Processing in Agriculture

Pub Date : 2025-03-01 DOI: 10.1016/j.inpa.2024.03.001

Gabriel S. Vieira , Afonso U. Fonseca , Naiane Maria de Sousa , Julio C. Ferreira , Juliana Paula Felix , Christian Dias Cabacinha , Fabrizzio Soares

As an essential component of the architecture of a plant, leaves are crucial to sustaining decision-making in cultivars and effectively support agricultural processes. When the leaf area is constantly monitored, a plant’s health and productive capacity can be assessed to foment proactive and reactive strategies. Because of that, one of the most critical tasks in agricultural processes is estimating foliar damage. In this sense, we present an automatic method to estimate leaf stress caused by insect herbivory, including damage in border regions. As a novelty, we present a method with well-defined processing steps suitable for numerical analysis and visual inspection of defoliation severity. We describe the proposed method and evaluate its performance concerning 12 different plant species. Experimental results show high assertiveness in estimating leaf area loss with a concordance correlation coefficient of 0.98 for grape, soybean, potato, and strawberry leaves. A classic pattern recognition approach, named template matching, is at the core of the method whose performance is compared to cutting-edge techniques. Results demonstrated that the method achieves foliar damage quantification with precision comparable to deep learning models. The code prepared by the authors is publicly available.

作为植物结构的重要组成部分，叶片对维持品种决策和有效支持农业进程至关重要。当不断监测叶面积时，可以评估植物的健康和生产能力，以促进主动和被动策略。正因为如此，农业过程中最关键的任务之一是估计叶面损害。在这个意义上，我们提出了一种自动估计昆虫食草性叶片胁迫的方法，包括边界地区的损害。作为一种新颖的方法，我们提出了一种具有明确定义的处理步骤的方法，适用于数值分析和落叶严重程度的目视检查。我们描述了所提出的方法，并评估了其在12种不同植物物种上的性能。实验结果表明，葡萄、大豆、马铃薯和草莓叶片的叶面积损失估计具有较高的自信，一致性相关系数为0.98。该方法的核心是一种经典的模式识别方法，即模板匹配方法，其性能与前沿技术相比较。结果表明，该方法达到了与深度学习模型相当的精度。作者编写的代码是公开的。

{"title":"An automatic method for estimating insect defoliation with visual highlights of consumed leaf tissue regions","authors":"Gabriel S. Vieira , Afonso U. Fonseca , Naiane Maria de Sousa , Julio C. Ferreira , Juliana Paula Felix , Christian Dias Cabacinha , Fabrizzio Soares","doi":"10.1016/j.inpa.2024.03.001","DOIUrl":"10.1016/j.inpa.2024.03.001","url":null,"abstract":"<div><div>As an essential component of the architecture of a plant, leaves are crucial to sustaining decision-making in cultivars and effectively support agricultural processes. When the leaf area is constantly monitored, a plant’s health and productive capacity can be assessed to foment proactive and reactive strategies. Because of that, one of the most critical tasks in agricultural processes is estimating foliar damage. In this sense, we present an automatic method to estimate leaf stress caused by insect herbivory, including damage in border regions. As a novelty, we present a method with well-defined processing steps suitable for numerical analysis and visual inspection of defoliation severity. We describe the proposed method and evaluate its performance concerning 12 different plant species. Experimental results show high assertiveness in estimating leaf area loss with a concordance correlation coefficient of 0.98 for grape, soybean, potato, and strawberry leaves. A classic pattern recognition approach, named template matching, is at the core of the method whose performance is compared to cutting-edge techniques. Results demonstrated that the method achieves foliar damage quantification with precision comparable to deep learning models. The code prepared by the authors is publicly available.</div></div>","PeriodicalId":53443,"journal":{"name":"Information Processing in Agriculture","volume":"12 1","pages":"Pages 40-53"},"PeriodicalIF":7.7,"publicationDate":"2025-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140084130","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A deep learning framework for prediction of crop yield in Australia under the impact of climate change 预测气候变化影响下澳大利亚作物产量的深度学习框架

IF 7.7 Q1 AGRICULTURE, MULTIDISCIPLINARY

Information Processing in Agriculture

Pub Date : 2025-03-01 DOI: 10.1016/j.inpa.2024.04.004

Haydar Demirhan

Accurate prediction of crop yields is essential to ensure food security. In this study, a new deep neural networks framework is developed to predict crop yields in Australia, considering the impact of climate change, fertilizer use, and crop area. It is implemented for oats, corn, rice, and wheat crops, and its forecasting performance is benchmarked against five statistical and machine learning methods. All the software codes for the implementation of the proposed framework are freely available. The proposed framework shows the highest forecasting performance for all the considered crop types. It provides 23%, 38%, 39%, and 40% lower average mean absolute error than the benchmark methods for oat, corn, rice, and wheat crops, respectively. The reductions in average root mean squared error are 19%, 25%, 37%, and 29% over the benchmark methods. Then, it is used to predict yields of the considered crops in Australia towards 2025 under six different climate change scenarios. It is observed that although climate change has some boosting impact on crop yield, it is not sustainable to meet the demand. However, it is possible to keep crop yields rising while mitigating climate change.

准确预测作物产量对确保粮食安全至关重要。在这项研究中，我们开发了一个新的深度神经网络框架来预测澳大利亚的作物产量，同时考虑了气候变化、肥料使用和作物面积的影响。它适用于燕麦、玉米、大米和小麦作物，其预测性能是根据五种统计和机器学习方法进行基准测试的。实现所提议的框架的所有软件代码都是免费提供的。建议的框架对所有考虑的作物类型显示出最高的预测性能。与燕麦、玉米、水稻和小麦的基准方法相比，该方法的平均绝对误差分别降低了23%、38%、39%和40%。与基准方法相比，平均均方根误差降低了19%、25%、37%和29%。然后，它被用来预测在六种不同的气候变化情景下，到2025年澳大利亚所考虑的作物的产量。我们观察到，气候变化虽然对作物产量有一定的促进作用，但满足需求是不可持续的。然而，在减缓气候变化的同时保持作物产量的增长是可能的。

{"title":"A deep learning framework for prediction of crop yield in Australia under the impact of climate change","authors":"Haydar Demirhan","doi":"10.1016/j.inpa.2024.04.004","DOIUrl":"10.1016/j.inpa.2024.04.004","url":null,"abstract":"<div><div>Accurate prediction of crop yields is essential to ensure food security. In this study, a new deep neural networks framework is developed to predict crop yields in Australia, considering the impact of climate change, fertilizer use, and crop area. It is implemented for oats, corn, rice, and wheat crops, and its forecasting performance is benchmarked against five statistical and machine learning methods. All the software codes for the implementation of the proposed framework are freely available. The proposed framework shows the highest forecasting performance for all the considered crop types. It provides 23%, 38%, 39%, and 40% lower average mean absolute error than the benchmark methods for oat, corn, rice, and wheat crops, respectively. The reductions in average root mean squared error are 19%, 25%, 37%, and 29% over the benchmark methods. Then, it is used to predict yields of the considered crops in Australia towards 2025 under six different climate change scenarios. It is observed that although climate change has some boosting impact on crop yield, it is not sustainable to meet the demand. However, it is possible to keep crop yields rising while mitigating climate change.</div></div>","PeriodicalId":53443,"journal":{"name":"Information Processing in Agriculture","volume":"12 1","pages":"Pages 125-138"},"PeriodicalIF":7.7,"publicationDate":"2025-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140761172","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Technologies, Protocols, and applications of Internet of Things in greenhouse Farming: A survey of recent advances 温室种植中的物联网技术、协议和应用：最新进展概览

IF 7.7 Q1 AGRICULTURE, MULTIDISCIPLINARY

Information Processing in Agriculture

Pub Date : 2025-03-01 DOI: 10.1016/j.inpa.2024.04.002

Khalid M. Hosny , Walaa M. El-Hady , Farid M. Samy

Greenhouse farming is considered one of the precision and sustainable forms of smart agriculture. Although greenhouse gases can support off-season crops inside the indoor environment, monitoring, controlling, and managing crop parameters at greenhouse farms more precisely and securely is necessary, even in harsh climate regions. The evolving Internet of Things (IoT) technologies, including smart sensors, devices, network topologies, big data analytics, and intelligent decision-making, are thought to be the solution for automating greenhouse farming parameters like internal atmosphere control, irrigation control, crop growth monitoring, and so on. This paper introduces a comprehensive survey of recent advances in IoT-based greenhouse farming. We summarize the related review articles. The classification of greenhouse farming based on IoT (smart greenhouse, hydroponics greenhouse, and vertical farming) is introduced. Also, we present a detailed architecture for the components of greenhouse agriculture applications based on IoT, including physical devices, communication protocols, and cloud/fog computing technologies. We also present a classification of IoT applications of greenhouse farming, including monitoring, controlling, tracking, and predicting. Furthermore, we present the technical and resource management challenges for optimal greenhouse farming. Moreover, countries already applying IoT in greenhouse farming have been presented. Lastly, future suggestions related to IoT-based greenhouse farming have been introduced.

温室农业被认为是智能农业的一种精确和可持续的形式。尽管温室气体可以在室内环境中支持淡季作物，但更精确、更安全地监测、控制和管理温室农场的作物参数是必要的，即使在气候恶劣的地区也是如此。不断发展的物联网（IoT）技术，包括智能传感器、设备、网络拓扑、大数据分析和智能决策，被认为是自动化温室农业参数的解决方案，如内部大气控制、灌溉控制、作物生长监测等。本文全面介绍了物联网温室农业的最新进展。我们对相关的综述文章进行了总结。介绍了基于物联网的温室农业分类（智能温室、水培温室、垂直农业）。此外，我们还提出了基于物联网的温室农业应用组件的详细架构，包括物理设备，通信协议和云/雾计算技术。我们还对温室农业的物联网应用进行了分类，包括监测、控制、跟踪和预测。此外，我们提出了优化温室农业的技术和资源管理挑战。此外，还介绍了已经在温室农业中应用物联网的国家。最后，对未来物联网温室农业的发展提出了建议。

{"title":"Technologies, Protocols, and applications of Internet of Things in greenhouse Farming: A survey of recent advances","authors":"Khalid M. Hosny , Walaa M. El-Hady , Farid M. Samy","doi":"10.1016/j.inpa.2024.04.002","DOIUrl":"10.1016/j.inpa.2024.04.002","url":null,"abstract":"<div><div>Greenhouse farming is considered one of the precision and sustainable forms of smart agriculture. Although greenhouse gases can support off-season crops inside the indoor environment, monitoring, controlling, and managing crop parameters at greenhouse farms more precisely and securely is necessary, even in harsh climate regions. The evolving Internet of Things (IoT) technologies, including smart sensors, devices, network topologies, big data analytics, and intelligent decision-making, are thought to be the solution for automating greenhouse farming parameters like internal atmosphere control, irrigation control, crop growth monitoring, and so on. This paper introduces a comprehensive survey of recent advances in IoT-based greenhouse farming. We summarize the related review articles. The classification of greenhouse farming based on IoT (smart greenhouse, hydroponics greenhouse, and vertical farming) is introduced. Also, we present a detailed architecture for the components of greenhouse agriculture applications based on IoT, including physical devices, communication protocols, and cloud/fog computing technologies. We also present a classification of IoT applications of greenhouse farming, including monitoring, controlling, tracking, and predicting. Furthermore, we present the technical and resource management challenges for optimal greenhouse farming. Moreover, countries already applying IoT in greenhouse farming have been presented. Lastly, future suggestions related to IoT-based greenhouse farming have been introduced.</div></div>","PeriodicalId":53443,"journal":{"name":"Information Processing in Agriculture","volume":"12 1","pages":"Pages 91-111"},"PeriodicalIF":7.7,"publicationDate":"2025-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140773497","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Society 5.0 enabled agriculture: Drivers, enabling technologies, architectures, opportunities, and challenges 社会 5.0 支持农业：驱动因素、使能技术、架构、机遇和挑战

IF 7.7 Q1 AGRICULTURE, MULTIDISCIPLINARY

Information Processing in Agriculture

Pub Date : 2025-03-01 DOI: 10.1016/j.inpa.2024.04.003

Kossi Dodzi Bissadu, Salleh Sonko, Gahangir Hossain

The existing agriculture practices faced many challenges and fail to address some of the most critical needs of the growing population. Food insecurity, high initial cost of smart farming, severe farm labor shortage worldwide, economic, social, and political crises related to famines, poverty, climate change, and the technology focus of Agriculture 4.0 calls for rethinking the agriculture paradigm. Moreover, the idea of Society 5.0 promoted by Japanese government triggered many position reactions from policymakers, governments, private institutions, academicians, and researchers. The idea of human centered society where individuals live their lives to the fullest with shared vision of happiness, social harmony, sustainability, and resilience recently caught scholars’ attention. Several researchers investigated the society 5.0 and its critical components including Agriculture 5.0. Agriculture 5.0 not only could be leveraged to address many existing issues, but could become a major driving force for achieving Society 5.0’s goals. This paper follows a systematic literature review approach to investigate the major drivers, enabling cutting-edge technologies, various opportunities and challenges for developing, adopting, and implementation Agriculture 5.0. It also highlighted the overall and holistic architectural framework based on 12 layers of Agriculture 5.0 paradigm. Though Agriculture 5.0 is promising with many opportunities, such as creating new job opportunities for young generations, and boosting mass customization, it will face many potential challenges. Some challenges include cybersecurity and privacy issues, difficulties for an effective legal, regulatory and compliance system due to high automation and mass personalization, standardization issues, and adapting agricultural production strategies and models to constantly changing customer preferences.

现有的农业实践面临许多挑战，无法满足不断增长的人口的一些最关键的需求。粮食不安全、智能农业的高初始成本、全球范围内严重的农业劳动力短缺、与饥荒、贫困、气候变化相关的经济、社会和政治危机以及农业4.0的技术重点要求我们重新思考农业范式。此外，日本政府提出的社会5.0理念引发了政策制定者、政府、私人机构、学者和研究人员的许多立场反应。最近，以人为中心的社会概念引起了学者们的关注，在这个社会中，每个人都有对幸福、社会和谐、可持续性和弹性的共同愿景，从而充分享受自己的生活。一些研究人员调查了社会5.0及其关键组成部分，包括农业5.0。农业5.0不仅可以用来解决许多现有的问题，而且可以成为实现社会5.0目标的主要推动力。本文采用系统的文献综述方法，研究了农业5.0的主要驱动因素、前沿技术、发展、采用和实施农业5.0的各种机遇和挑战。强调了基于12层农业5.0范式的整体整体架构框架。“农业5.0”虽然为年轻一代创造新的就业机会、促进大规模定制等带来了很多机会，但也面临着许多潜在的挑战。一些挑战包括网络安全和隐私问题，由于高度自动化和大规模个性化，难以建立有效的法律、监管和合规系统，标准化问题，以及使农业生产战略和模式适应不断变化的客户偏好。

{"title":"Society 5.0 enabled agriculture: Drivers, enabling technologies, architectures, opportunities, and challenges","authors":"Kossi Dodzi Bissadu, Salleh Sonko, Gahangir Hossain","doi":"10.1016/j.inpa.2024.04.003","DOIUrl":"10.1016/j.inpa.2024.04.003","url":null,"abstract":"<div><div>The existing agriculture practices faced many challenges and fail to address some of the most critical needs of the growing population. Food insecurity, high initial cost of smart farming, severe farm labor shortage worldwide, economic, social, and political crises related to famines, poverty, climate change, and the technology focus of Agriculture 4.0 calls for rethinking the agriculture paradigm. Moreover, the idea of Society 5.0 promoted by Japanese government triggered many position reactions from policymakers, governments, private institutions, academicians, and researchers. The idea of human centered society where individuals live their lives to the fullest with shared vision of happiness, social harmony, sustainability, and resilience recently caught scholars’ attention. Several researchers investigated the society 5.0 and its critical components including Agriculture 5.0. Agriculture 5.0 not only could be leveraged to address many existing issues, but could become a major driving force for achieving Society 5.0’s goals. This paper follows a systematic literature review approach to investigate the major drivers, enabling cutting-edge technologies, various opportunities and challenges for developing, adopting, and implementation Agriculture 5.0. It also highlighted the overall and holistic architectural framework based on 12 layers of Agriculture 5.0 paradigm. Though Agriculture 5.0 is promising with many opportunities, such as creating new job opportunities for young generations, and boosting mass customization, it will face many potential challenges. Some challenges include cybersecurity and privacy issues, difficulties for an effective legal, regulatory and compliance system due to high automation and mass personalization, standardization issues, and adapting agricultural production strategies and models to constantly changing customer preferences.</div></div>","PeriodicalId":53443,"journal":{"name":"Information Processing in Agriculture","volume":"12 1","pages":"Pages 112-124"},"PeriodicalIF":7.7,"publicationDate":"2025-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140777764","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Few-shot cow identification via meta-learning 通过元学习进行奶牛识别

IF 7.7 Q1 AGRICULTURE, MULTIDISCIPLINARY

Information Processing in Agriculture

Pub Date : 2025-03-01 DOI: 10.1016/j.inpa.2024.04.001

Xingshi Xu, Yunfei Wang, Yuying Shang, Guangyuan Yang, Zhixin Hua, Zheng Wang, Huaibo Song

Cow identification is a prerequisite for precision livestock farming. Biometric-based methods have made significant progress in cow identification. However, substantial labelling costs and frequent identification task changes are still hamper model application. In this work, a novel method called “MFCI” was proposed to achieve accurate cow identification under few-shot and task-changing conditions. Specifically, the proposed method comprises two components: cow location and cow identification. First, an improved YOLOv5n with Ghost module was adopted to quickly detect cow locations in images. Then, the Model-Agnostic Meta-Learning (MAML) framework was introduced for accurate identification under few-shot conditions and for fast adaptation to frequent changes in individual cows. Moreover, an autoencoder was adopted to allow Base-Learner learn more generalized features by combining both supervised and unsupervised approaches. The experimental results showed that the proposed cow location model achieved a mAP of 99.5 %. The proposed cow identification model attained an accuracy of 90.43 % with only five samples per cow for 20 cows, outperforming other state-of-the-art methods. The results demonstrate the broad applicability and significant value of the proposed method.

牛的鉴定是精确畜牧业的先决条件。基于生物特征的方法在奶牛识别方面取得了重大进展。然而，大量的标签成本和频繁的识别任务变化仍然阻碍了模型的应用。在这项工作中，提出了一种称为“MFCI”的新方法，以实现在少量射击和任务变化条件下准确识别奶牛。具体来说，该方法包括两个部分：奶牛定位和奶牛识别。首先，采用改进的带有Ghost模块的YOLOv5n快速检测图像中的奶牛位置。然后，引入了模型不可知元学习（Model-Agnostic Meta-Learning， MAML）框架，以便在少量条件下准确识别，并快速适应奶牛个体的频繁变化。此外，采用了自动编码器，通过结合监督和无监督方法，使Base-Learner能够学习更多的广义特征。实验结果表明，所提出的奶牛定位模型的mAP值达到了99.5%。所提出的奶牛识别模型在20头奶牛中每头奶牛只有5个样本，准确率达到90.43%，优于其他最先进的方法。结果表明，该方法具有广泛的适用性和重要的应用价值。

{"title":"Few-shot cow identification via meta-learning","authors":"Xingshi Xu, Yunfei Wang, Yuying Shang, Guangyuan Yang, Zhixin Hua, Zheng Wang, Huaibo Song","doi":"10.1016/j.inpa.2024.04.001","DOIUrl":"10.1016/j.inpa.2024.04.001","url":null,"abstract":"<div><div>Cow identification is a prerequisite for precision livestock farming. Biometric-based methods have made significant progress in cow identification. However, substantial labelling costs and frequent identification task changes are still hamper model application. In this work, a novel method called “MFCI” was proposed to achieve accurate cow identification under few-shot and task-changing conditions. Specifically, the proposed method comprises two components: cow location and cow identification. First, an improved YOLOv5n with Ghost module was adopted to quickly detect cow locations in images. Then, the Model-Agnostic Meta-Learning (MAML) framework was introduced for accurate identification under few-shot conditions and for fast adaptation to frequent changes in individual cows. Moreover, an autoencoder was adopted to allow Base-Learner learn more generalized features by combining both supervised and unsupervised approaches. The experimental results showed that the proposed cow location model achieved a mAP of 99.5 %. The proposed cow identification model attained an accuracy of 90.43 % with only five samples per cow for 20 cows, outperforming other state-of-the-art methods. The results demonstrate the broad applicability and significant value of the proposed method.</div></div>","PeriodicalId":53443,"journal":{"name":"Information Processing in Agriculture","volume":"12 1","pages":"Pages 80-90"},"PeriodicalIF":7.7,"publicationDate":"2025-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140767863","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Supervised and unsupervised machine learning approaches for prediction and geographical discrimination of Iranian saffron ecotypes based on flower-related and phytochemical attributes 基于花卉相关属性和植物化学属性的伊朗藏红花生态型预测和地理鉴别的有监督和无监督机器学习方法

IF 7.7 Q1 AGRICULTURE, MULTIDISCIPLINARY

Information Processing in Agriculture

Pub Date : 2025-03-01 DOI: 10.1016/j.inpa.2023.12.002

Seid Mohammad Alavi-Siney , Jalal Saba , Alireza Fotuhi Siahpirani , Jaber Nasiri

A two-year field experiment (2014–2016; Zanjan, Iran) was conducted to monitor potential diversity pattern and adaptability power among 18 Iranian saffron ecotypes under Zanjan climatological conditions using seven flower-related and three qualitative traits (crocin, picrocrocin, and safranal, determined by UV–visible spectra), and analyzed by supervised and unsupervised approaches. A range of variability was recorded among the ecotypes, and despite some exceptions, overall, saffron corms produced higher amounts of studied features across the second year. The Feizabad ecotype was recommended to acquire maximum qualitative criteria (category I; based on ISO Normative 3632 grading system), while for flower-related parameters several ecotypes (e.g., Ghaien, Bardeskan, Torbat-Jam, and Gonabad) could be applied for Zanjan climatological conditions. Based on the results of Leave-One-Out Cross-Validation (LOOCV), various prediction values were computed for all 10 classifiers of LDA, QDA, FDA, MDA, RDA, Naive Bayes, Decision Tree, Linear SVM, Radial SVM, and Random Forest in terms of accuracy, sensitivity and specificity parameters. Among which, Random Forest and LDA with the values of 0.91 and 0.78 possessed the highest and the lowest amounts of accuracy, respectively. Finally, considering the highest accuracy value of the superior classification model of Random forest, both feature subsets of “FFW, FDW, Picrocrocin, Safranal, and Crocin” and “SFW, FDW, Picrocrocin, Safranal, and Crocin” were nominated as the most powerful elements (comparing to the remaining 1021 feature subsets) to make accurate discrimination between Khorasan and non-Khorasan saffron ecotypes. The results, overall, revealed that saffron ecotypes followed different responses under Zanjan climatological circumstances, and Random Forest is more suitable for accurately predicting saffron corms from different provenances.

为期两年的野外实验(2014-2016；利用7个花相关性状和3个质量性状（藏红花素、微番红花素和番红花素，由紫外可见光谱测定），对伊朗18个藏红花生态型在赞詹气候条件下的潜在多样性格局和适应能力进行了监测，并采用监督和非监督方法进行了分析。在生态型中记录了一系列的变化，尽管有一些例外，总的来说，藏红花球茎在第二年产生了更多的研究特征。Feizabad生态型被推荐获得最高的质量标准(第一类；基于ISO标准3632分级系统)，而对于与花相关的参数，几个生态型（例如，Ghaien, Bardeskan， Torbat-Jam和Gonabad）可以适用于赞詹的气候条件。基于LOOCV交叉验证结果，计算LDA、QDA、FDA、MDA、RDA、朴素贝叶斯、决策树、线性支持向量机、径向支持向量机和随机森林10种分类器在准确率、灵敏度和特异性参数方面的预测值。其中Random Forest和LDA的准确率最高，分别为0.91和0.78。最后，考虑到随机森林优势分类模型的最高准确率值，将“FFW， FDW, Picrocrocin, Safranal, and Crocin”和“SFW， FDW, Picrocrocin, Safranal, and Crocin”两个特征子集提名为准确区分呼罗珊和非呼罗珊藏红花生态类型的最强大元素（与其余1021个特征子集相比）。结果表明，在赞詹气候条件下，藏红花生态型表现出不同的响应，随机森林更适合于对不同种源藏红花球茎的准确预测。

{"title":"Supervised and unsupervised machine learning approaches for prediction and geographical discrimination of Iranian saffron ecotypes based on flower-related and phytochemical attributes","authors":"Seid Mohammad Alavi-Siney , Jalal Saba , Alireza Fotuhi Siahpirani , Jaber Nasiri","doi":"10.1016/j.inpa.2023.12.002","DOIUrl":"10.1016/j.inpa.2023.12.002","url":null,"abstract":"<div><div>A two-year field experiment (2014–2016; Zanjan, Iran) was conducted to monitor potential diversity pattern and adaptability power among 18 Iranian saffron ecotypes under Zanjan climatological conditions using seven flower-related and three qualitative traits (crocin, picrocrocin, and safranal, determined by UV–visible spectra), and analyzed by supervised and unsupervised approaches. A range of variability was recorded among the ecotypes, and despite some exceptions, overall, saffron corms produced higher amounts of studied features across the second year. The Feizabad ecotype was recommended to acquire maximum qualitative criteria (category I; based on ISO Normative 3632 grading system), while for flower-related parameters several ecotypes (e.g., Ghaien, Bardeskan, Torbat-Jam, and Gonabad) could be applied for Zanjan climatological conditions. Based on the results of Leave-One-Out Cross-Validation (LOOCV), various prediction values were computed for all 10 classifiers of LDA, QDA, FDA, MDA, RDA, Naive Bayes, Decision Tree, Linear SVM, Radial SVM, and Random Forest in terms of accuracy, sensitivity and specificity parameters. Among which, Random Forest and LDA with the values of 0.91 and 0.78 possessed the highest and the lowest amounts of accuracy, respectively. Finally, considering the highest accuracy value of the superior classification model of Random forest, both feature subsets of “FFW, FDW, Picrocrocin, Safranal, and Crocin” and “SFW, FDW, Picrocrocin, Safranal, and Crocin” were nominated as the most powerful elements (comparing to the remaining 1021 feature subsets) to make accurate discrimination between Khorasan and non-Khorasan saffron ecotypes. The results, overall, revealed that saffron ecotypes followed different responses under Zanjan climatological circumstances, and Random Forest is more suitable for accurately predicting saffron corms from different provenances.</div></div>","PeriodicalId":53443,"journal":{"name":"Information Processing in Agriculture","volume":"12 1","pages":"Pages 1-16"},"PeriodicalIF":7.7,"publicationDate":"2025-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"138609531","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0