[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"item-186":3,"related-186":62},{"id":4,"title":5,"url":6,"summary":7,"summary_zh":8,"content":9,"source_name":10,"source_url":8,"published_at":11,"category":12,"cover_url":8,"hotness":13,"is_selected":14,"score":15,"score_detail":16,"sources":23,"tags":25,"primary_topic":29,"secondary_topics":35,"classification":36,"search_phrases":55,"slug":58,"view_count":59,"doi":60,"paper":8,"created_at":61},186,"基于影像组学与可解释机器学习的香蕉病害快速识别","https:\u002F\u002Fwww.frontiersin.org\u002Fjournals\u002Fartificial-intelligence\u002Farticles\u002F10.3389\u002Ffrai.2026.1861135\u002Ffull","原文：Radiomics-driven and explainable machine learning for rapid characterization of Fusarium wilt and Black Sigatoka in banana crops。研究于2026年7月28日发表，聚焦利用影像组学和可解释机器学习快速表征香蕉枯萎病与黑斑病，属于作物病害智能识别方向。当前公开检索摘要未呈现完整实验细节，故不对模型精度作额外推断。",null,"## Abstract\n\n**Introduction:**\n\nBanana production is increasingly threatened by fungal diseases such as Fusarium wilt and Black Sigatoka, posing severe risks to food security and agricultural economies. Recent image-based approaches using deep learning have shown high predictive capacity for plant disease recognition; however, their limited transparency, calibration uncertainty, and sensitivity to domain shifts can restrict their use in decision-support workflows that require auditability.\n\n**Methods:**\n\nThis study proposes an interpretable and calibrated Artificial Intelligence framework for multiclass banana disease-pattern characterization based on radiomic feature analysis of RGB leaf images. Radiomic features were extracted from HSV-segmented banana leaf regions, resulting in a dataset of 14,763 samples characterized by 103 quantitative descriptors and labeled as Healthy, Sigatoka, or Fusarium wilt race 1.\n\n**Results:**\n\nAmong the evaluated radiomics classifiers, the calibrated Random Forest achieved accuracy = 0.85, balanced accuracy = 0.84, macro-F1 = 0.84, and macro ROC-AUC OvR = 0.95 on the held-out test set. Bootstrap analysis yielded 95% confidence intervals of [0.8406, 0.8691] for accuracy and [0.8306, 0.8606] for balanced accuracy. Three deep learning baselines trained on the same partition achieved higher predictive performance: MobileNetV3 with accuracy = 0.96, macro-F1 = 0.95, and macro ROC-AUC OvR = 0.97; EfficientNet with accuracy = 0.96, macro-F1 = 0.97, and macro ROC-AUC OvR = 0.96; and ResNet-18 with accuracy = 0.96, macro-F1 = 0.96, and macro ROC-AUC OvR = 0.96.\n\n**Discussion:**\n\nThe CNNs produced strong classification performance on the evaluated repositories, and the radiomics approach demonstrated to be a complementary interpretable and explainable calibrated reference model. SHAP, LIME, permutation importance, accumulated local effects, calibration curves, and Brier score decomposition supported feature-level inspection of the final model.\n\n## 1 Introduction\n\nBanana (_Musa_ spp.) is one of the most important staple and export crops worldwide, playing a critical role in food security and socioeconomic stability across tropical and subtropical regions (). Ecuador is one of the world's leading exporters of bananas, and the crop represents a strategic pillar of its national agricultural economy (). However, banana production is increasingly threatened by fungal diseases that compromise yield, fruit quality, and plantation sustainability, most notably Fusarium wilt and Black Sigatoka (). The growing pressure exerted by these diseases has intensified the need for rapid, early, accurate, scalable, and interpretable diagnostic tools capable of supporting timely phytosanitary interventions.\n\nFusarium wilt of banana, caused by _Fusarium oxysporum_ f. sp. _cubense_, is considered one of the most destructive plant diseases globally because of its soilborne nature, long persistence, and the lack of effective chemical control (). Of particular concern is Tropical Race 4 (TR4), which has spread across Asia, Africa, and Latin America in recent years (). In Ecuador, recent phytosanitary alerts and institutional containment actions have reinforced the need for surveillance technologies that support risk prioritization, field inspection, and confirmatory laboratory workflows as stated by the Ecuadorian agricultural authority . Although these technologies are not intended to replace official diagnosis, they can help to enhance disease monitoring, the recognition of disease patterns, and prioritize suspect cases for expert inspection and laboratory testing (; ).\n\nBlack Sigatoka, caused by _Pseudocercospora fijiensis_, represents another major fungal threat, leading to leaf necrosis, reduced photosynthetic capacity, and yield losses if not properly managed (). Unlike Fusarium wilt, Sigatoka symptoms are primarily foliar and visually detectable; however, symptom progression varies with environmental conditions, disease stage, cultivar, and management practice, making manual recognition subjective and error prone. In addition, visual overlap among disease-related leaf alterations may lead to delayed response or incorrect operational actions (; ).\n\nTraditional disease identification in banana plantations relies heavily on expert visual inspection and laboratory-based confirmation. While effective, these approaches are time consuming, costly, and difficult to deploy continuously across large agricultural areas, particularly when diseases such as Foc TR4 require coordinated biosecurity, surveillance, and confirmatory response workflows (; ). Artificial Intelligence (AI) and Machine Learning (ML) have therefore emerged as tools for automated analysis of visual and spectral data in plant disease detection (; ). Deep learning (DL) architectures, including convolutional neural networks (CNNs), MobileNets, Vision Transformers, and emerging foundation or vision-language models, can achieve strong performance in plant disease recognition from RGB, UAV, multispectral, or hyperspectral imagery (; ; ). Nevertheless, high predictive accuracy alone is not sufficient for phytosanitary decision support: transparency, probability calibration, leakage-aware validation, interpretability, explainability and external testing are also required (; ; ).\n\nRadiomics is an approach for extracting quantitative descriptors that characterize intensity, shape, and texture patterns from images (). Originally developed in medical imaging, radiomics is conceptually aligned with image-based plant phenotyping because both aim to convert image patterns into quantitative, analyzable traits (). Texture descriptors derived from gray-level co-occurrence matrices (GLCM), gray-level run-length matrices (GLRLM), gray-level size-zone matrices (GLSZM), gray-level dependence matrices (GLDM), and neighboring gray-tone difference matrices (NGTDM) can capture lesion heterogeneity, tissue irregularity, and spatial organization; these feature families are widely used in standardized radiomics implementations such as PyRadiomics and follow IBSI-oriented definitions (; ). When combined with classical ML classifiers, radiomics enables models whose predictions can be inspected at the feature level using permutation importance, SHAP, LIME, and accumulated local effects (; ).\n\nThe methodological gap addressed in this study is not the absence of high-accuracy image classifiers. Rather, the gap lies in the limited availability of calibrated, leakage-aware, and feature-level interpretable baselines for multiclass banana disease-pattern characterization involving H ealthy, S igatoka, and F usarium wilt categories. This study evaluates a radiomics-based ML framework together with three CNN baselines (MobileNetV3, EfficientNet, and ResNet-18), to quantify the accuracy–interpretability trade-off under the same data partition. The radiomics model is positioned as a complementary, interpretable and auditable reference model, whereas the CNN baselines establish strong performance image-based benchmarks.\n\nThe remainder of this paper is organized as follows. Section 2 reviews recent AI-based approaches for plant disease characterization, including DL, interpretable models, calibration, and leakage-aware validation. Section 3 describes the data sources, leakage-aware partitioning, HSV-based segmentation, radiomic feature extraction, feature preprocessing, classical classifiers, CNN baselines, probability calibration, and statistical analyses. Section 4 presents the comparative results, bootstrap confidence intervals, McNemar testing, CNN baseline comparison, calibration assessment, and explainability analyses. Section 5 discusses the biological plausibility, operational implications, domain-shift limitations, and deployment constraints. Section 6 concludes with future directions for Ecuadorian field validation and hybrid interpretable DL systems.\n\n## 2 Related work\n\n### 2.1 AI-based approaches for plant disease characterization\n\nFusarium wilt, in particular TR4, has attracted increasing research attention because of its rapid spread and impact on banana production (). Existing characterization strategies focus primarily on molecular diagnostics, soil assays, and expert visual inspection (). While these methods are reliable, they are not easily scalable for continuous field monitoring. Recent work on Fusarium TR4 in Latin America and the Caribbean has emphasized that effective management requires surveillance, biosecurity, containment, resistant cultivars, and coordinated regional response strategies (). In addition, the socioeconomic effects of TR4 on banana producers in Latin America further reinforce the need for scalable monitoring and early-warning tools (). AI-driven approaches for Fusarium wilt characterization remain more limited than those for foliar diseases, and many image-based studies do not explicitly address differentiation between Fusarium wilt and other diseases with overlapping visual characteristics.\n\nBlack Sigatoka has been more extensively studied using image-based techniques, but many studies address binary classification or disease severity estimation rather than multiclass differentiation among healthy leaves and multiple disease types. Recent studies using UAV-based multispectral imagery and DL have demonstrated the feasibility of plantation-level monitoring; however, these approaches remain sensitive to dataset scale, geographic variability, acquisition devices, and environmental conditions (). These limitations are relevant for banana disease recognition because cultivar diversity, field management, illumination, canopy structure, and regional disease pressure can affect the visual appearance of symptoms and the generalization of image-based models.\n\nAI techniques have been explored for automating the recognition of disease patterns, mainly based on computer vision and using ML techniques. Early approaches focused on handcrafted features combined with classical classifiers, and recent reviews of plant disease detection emphasize that DL has become a dominant modeling paradigm because of its capacity to learn discriminative visual features directly from images (; ). Compact CNNs such as MobileNet variants are particularly relevant for agricultural applications because they reduce parameter count and computational requirements relative to larger architectures, whereas EfficientNet and ResNet families remain common image-classification benchmarks. At the same time, recent literature highlights continuing challenges in cross-domain generalization, benchmark design, data leakage, annotation quality, and interpretability (; ).\n\nCNN-based approaches have been applied to detect foliar diseases such as Black Sigatoka, using leaf images captured under natural illumination. Previous studies (; ) have reported high classification accuracy using architectures such as VGG, ResNet, MobileNet, and related CNN variants, showing the potential of DL for large-scale disease monitoring (; ). UAV-based and multispectral imaging combined with deep neural networks have further extended these capabilities to plantation-level analysis, including Black Sigatoka monitoring in banana crops (; ). However, these models often require sufficiently large and well-annotated datasets, and their performance may be sensitive to domain shifts caused by illumination, background variability, acquisition devices, geographic conditions, and canopy structure. In addition, many DL models operate through latent feature representations, which can limit direct feature-level interpretability and complicate adoption in decision-support or regulatory contexts (; ).\n\nParallel to DL developments, classical ML methods using handcrafted or radiomic descriptors remain relevant, particularly in scenarios where calibration, computational constraints, auditability, or explainability are important. Texture-based descriptors, including Haralick features derived from gray-level co-occurrence matrices (GLCM), have been widely used to characterize plant disease symptoms because they capture spatial heterogeneity and lesion structure (; ; ). These approaches are generally more transparent at the feature level than end-to-end DL models, although they may provide lower raw predictive accuracy. This motivates a direct comparison between calibrated radiomics models and CNN baselines under the same data partition, allowing the accuracy–interpretability trade-off to be quantified empirically.\n\n### 2.2 Radiomics, calibration, and explainability in agricultural AI\n\nRadiomics extends traditional texture analysis by extracting a large set of quantitative descriptors that characterize intensity distributions, shape, spatial heterogeneity, and multiscale texture (). Originally developed in medical imaging, radiomics has recently been adopted in agricultural and plant phenotyping contexts, where subtle morphological and textural changes may reflect disease progression, stress responses, or tissue degradation (; ). These descriptors provide an intermediate representation between raw image pixels and model predictions, enabling quantitative characterization of visual patterns that may otherwise be difficult to inspect directly.\n\nRecent research has demonstrated the utility of radiomic and texture-based features, such as first-order statistics, shape descriptors, GLCM, GLRLM, GLSZM, GLDM, NGTDM, and wavelet-derived descriptors, for characterizing heterogeneous image patterns associated with plant stress and disease. Foundational texture descriptors (; ), including Haralick features derived from gray-level co-occurrence matrices, remain relevant for methodological definitions because they capture spatial heterogeneity and local gray-level relationships. In plant pathology, these feature families are useful because fungal infections can induce lesion coalescence, tissue irregularity, necrotic patterns, and non-uniform spatial organization, all of which can be represented through quantitative shape and texture descriptors.\n\nBeyond predictive accuracy, recent research also emphasizes the importance of model calibration and interpretability for deploying AI systems in high-stakes or decision-support domains, including agriculture (; ). Poorly calibrated models may produce overconfident predictions, leading to inappropriate prioritization, inspection, or management decisions. Techniques such as Isotonic Regression () and Platt scaling () have been proposed to improve probability reliability when predicted probabilities are used for risk-aware decision support (; ). In phytosanitary workflows, calibration is relevant because model outputs may be used to define inspection thresholds, prioritize uncertain cases, or communicate confidence to agronomists and plant-health authorities.\n\nExplainable AI (XAI) methods, including Permutation Importance, SHAP, LIME, and Accumulated Local Effects (ALE), are increasingly used to interpret model behavior and identify features associated with model predictions (; ; ). Recent agricultural AI literature has further emphasized the role of XAI in improving transparency, interpretability, and trust in AI-driven decision-making systems (; ). These techniques are particularly valuable in radiomics-based models because explanations can be linked to explicit image descriptors such as perimeter, sphericity, intensity percentiles, and texture non-uniformity. This feature-level interpretation distinguishes radiomics from end-to-end CNN models, where explanations often require post-hoc saliency or attribution methods over latent representations.\n\nRecent work on data splitting in biological ML has also shown that random or naïve partitions may inflate performance when related biological entities, duplicated observations, or highly similar samples appear in both training and test sets. () introduced DataSAIL as a leakage-reduced splitting strategy intended to enable more realistic evaluation of ML models in biological data, particularly under out-of-distribution settings. This issue is relevant for agricultural image datasets because images may share acquisition conditions, plant identity, plot identity, or visually similar backgrounds. Therefore, leakage-aware validation is necessary to avoid overly optimistic performance estimates and to better approximate deployment conditions.\n\nWhile DL approaches dominate recent studies on plant disease recognition, there remains a need for calibrated, leakage-aware, and feature-level interpretable frameworks tailored to multiclass banana disease-pattern characterization. Radiomics-based ML approaches provide a complementary reference model for this purpose as an auditable framework that allows probability calibration and inspection of quantitative descriptors associated with disease-related image patterns. In this study a joint evaluation of a calibrated radiomics model and CNN baselines under the same data partition is proposed.\n\n## 3 Materials and methods\n\n### 3.1 Data sources and leakage-aware partitions\n\nTwo publicly available banana leaf image repositories were used to construct a three-class dataset comprising H ealthy, S igatoka, and F usarium wilt race 1 (F usarium1) samples. Images corresponding to Fusarium wilt and a subset of Healthy leaves were obtained from the Banana Leaves Imagery Dataset collected in Tanzania (; ). Additional Healthy samples and Sigatoka images were sourced from the Banana Leaf Disease Dataset Bangladesh (). All images were RGB and organized in class-specific directories prior to preprocessing ([Figure 1](https:\u002F\u002Fwww.frontiersin.org\u002Fjournals\u002Fartificial-intelligence\u002Farticles\u002F10.3389\u002Ffrai.2026.1861135\u002Ffull#F1)).\n\nFigure 1\n\nFollowing image segmentation and radiomic feature extraction, the resulting dataset consisted of _N_ = 14, 763 radiomic samples. The class distribution after radiomic extraction was 4,866 samples for the H ealthy class, 5,767 samples for the S igatoka class, and 4,130 samples for the F usarium1 class ([Figure 2](https:\u002F\u002Fwww.frontiersin.org\u002Fjournals\u002Fartificial-intelligence\u002Farticles\u002F10.3389\u002Ffrai.2026.1861135\u002Ffull#F2)). This distribution reflects moderate class imbalance, motivating the use of balanced accuracy and macro-averaged F1-score throughout the study.\n\nFigure 2\n\nTo reduce the risk of information leakage, the experimental protocol used a leakage-aware partitioning strategy based on original image identity. A group identifier was constructed from the original image path\u002Fstem, and samples derived from the same source image identity were assigned to only one of the training, validation, or test subsets. The dataset was partitioned into training, validation, and test subsets with proportions of 70%, 15%, and 15%, respectively. The test subset was held out and remained frozen during model selection, hyperparameter tuning, feature preprocessing, and probability calibration.\n\nThe public repositories used in this study do not provide verified plant-level, plot-level, or farm-level identifiers. Therefore, the present validation should not be interpreted as a plant-level or plot-level external validation. A stricter plant\u002Fplot-level or similarity-aware split, such as DataSAIL-like splitting (), would be preferable when the corresponding metadata or image-similarity matrices become available.\n\n#### 3.1.1 Definition of experimental unit and radiomic samples\n\nIn this study, the experimental unit was defined at the image level. Each RGB image obtained from the public banana leaf repositories was associated with one class label according to the original dataset annotation: H ealthy, S igatoka, or F usarium1. After HSV-based segmentation, a leaf-level region of interest (ROI) was extracted from each image and used for radiomic feature computation. Thus, each radiomic sample corresponds to one image-derived observation represented by a 103-dimensional radiomic feature vector.\n\nThe final dataset contained 14,763 radiomic samples, corresponding to","Frontiers in Artificial Intelligence","2026-07-28T00:00:00Z","论文",10,false,60,{"impact":17,"substance":18,"depth":19,"authority":17,"freshness":20,"relevant":21,"comment":22},12,18,16,2,1,"论文提出可解释的香蕉病害识别方法，对农业信息化有参考价值，但时效性较低。",[24],{"name":10,"url":6},[26,27,28],"智慧农业","作物病害","可解释机器学习",{"code":30,"slug":31,"name":32,"path":33,"parent_code":34},"I2","agricultural-ai","农业AI与大模型","\u002Ftopic\u002Fdigital-agriculture\u002Fagricultural-ai","I",[],{"status":37,"locked":14,"version":38,"method":39,"evidence":40},"classified","2026-10-22.2","rules",{"candidates":41,"agricultural_evidence":44},[42,50],{"code":30,"score":43,"supported":44,"title":45,"abstract":47,"tags":49},15,true,[46],"机器学习",[46,48],"machine learning",[28],{"code":51,"score":21,"supported":14,"title":52,"abstract":53,"tags":54},"I1",[],[],[26],[56,57],"可解释机器学习 作物病害 智慧农业","可解释机器学习 作物病害","可解释机器学习作物病害智慧农业-186",7,"10.3389\u002Ffrai.2026.1861135\u002Ffull","2026-08-02T23:54:41.266361Z",{"total":63,"page":21,"page_size":63,"items":64},6,[65,128,207,252,306,365],{"id":66,"title":67,"url":68,"summary":69,"summary_zh":70,"content":8,"source_name":71,"source_url":68,"published_at":72,"category":12,"cover_url":8,"hotness":13,"is_selected":14,"score":73,"score_detail":74,"sources":78,"tags":80,"primary_topic":84,"secondary_topics":85,"classification":86,"search_phrases":105,"slug":108,"view_count":109,"doi":110,"paper":111,"created_at":127},5584,"Distributionally-Robust Evaluation of Computer Vision Agriculture Disease Classifiers Under Climate-Driven Pathogen Range Shifts: An Artificial Intelligence Worst-Case Generalization Bound","https:\u002F\u002Fdoi.org\u002F10.5281\u002Fzenodo.23544150","Computer vision classifiers for crop disease are evaluated almost exclusively on held-out samples drawn from the same distribution as their training data, yet climate-driven pathogen range shifts mean that deployment environments are systematically out of distribution. The framework treats the deployment distribution as an adversarial perturbation of the training distribution within a Wasserstein ball whose radius is calibrated to observed climate displacement. Using NASA POWER monthly reanalysis for Harare, Zimbabwe (2000-2023, 312 monthly observations), the analysis estimates a mean air temperature of 18.914 C, a standard deviation of 3.107 C, and a warming trend of 0.2001 C per decade. These estimates anchor the radius of the ambiguity set and the resulting bound. The bound decomposes worst-case risk into an empirical risk term, a Wasserstein radius term scaled by the Lipschitz constant of the classifier, and a concentration term. Under plausible Lipschitz constants for convolutional backbones, the radius term dominates the empirical risk term once the climate displacement exceeds a threshold that the Harare record already approaches. The contribution is a formal evaluation criterion that replaces the i. i. d. assumption with a climate-calibrated ambiguity set, together with the analytical machinery needed to compute it.","作物病害的计算机视觉分类器几乎完全在与训练数据同分布的留出样本上进行评估，然而气候驱动的病原体分布范围变化意味着部署环境系统性地处于分布之外。该框架将部署分布视为训练分布在Wasserstein球内的对抗性扰动，球的半径根据观测到的气候位移进行校准。利用NASA POWER对津巴布韦哈拉雷的月度再分析数据（2000—2023年，312个月度观测值），分析估计平均气温为18.914°C，标准差为3.107°C，每十年升温趋势为0.2001°C。这些估计值锚定了模糊集的半径及由此得到的界。该界将最坏情况风险分解为经验风险项、由分类器Lipschitz常数缩放后的Wasserstein半径项以及集中项。在卷积骨干网络合理的Lipschitz常数下，一旦气候位移超过哈拉雷记录已接近的阈值，半径项便主导经验风险项。本文的贡献在于提出了一种形式化评估准则，以气候校准的模糊集替代独立同分布（i.i.d.）假设，并提供了计算该准则所需的分析工具。","Zenodo (CERN European Organization for Nuclear Research)","2026-10-21T00:00:00Z",75,{"impact":17,"substance":75,"depth":18,"authority":76,"freshness":13,"relevant":21,"comment":77},22,13,"方法新颖、数据具体，但属方法论预印论文，产业影响有限，适合作为专业精选而非头条。",[79],{"name":71,"url":68},[26,81,82,27,83],"农业人工智能","计算机视觉","气候变化",{"code":30,"slug":31,"name":32,"path":33,"parent_code":34},[],{"status":37,"locked":14,"version":38,"method":39,"evidence":87},{"candidates":88,"agricultural_evidence":44},[89,96,101],{"code":30,"score":90,"supported":44,"title":91,"abstract":94,"tags":95},24,[92,93],"artificial intelligence","computer vision",[82,93],[81,82],{"code":97,"score":21,"supported":14,"title":98,"abstract":99,"tags":100},"H7",[],[],[83],{"code":51,"score":21,"supported":14,"title":102,"abstract":103,"tags":104},[],[],[26],[106,107],"NASA POWER 哈拉雷 气温","Wasserstein 分布鲁棒 病害分类","NASAPOWER哈拉雷气温-5584",0,"10.5281\u002Fzenodo.23544150",{"doi":110,"openalex_id":112,"authors":113,"venue":71,"cited_by_count":109,"oa_url":68,"card":120,"direction":124,"ingested_from":126},"W7223087892",[114,116,118],{"name":115,"orcid":8},"Kossi Afeli",{"name":117,"orcid":8},"Akouvi Amegashie",{"name":119,"orcid":8},"Komlan Sélom Dossou",{"tldr":121,"method":122,"finding":123,"direction":124,"opportunity":125},"提出气候校准的分布鲁棒评估准则，替代i.i.d.假设评价作物病害视觉分类器。","用Wasserstein球建模部署分布偏移，以NASA POWER哈拉雷气温数据","气候位移超阈值后，半径项主导经验风险，现有评估会高估模型泛化能力。","农业人工智能与决策模型","可将气候校准的鲁棒评估扩展到多作物多区域，并研究适应分布偏移的鲁棒训练方法。","openalex","2026-10-22T11:21:00.036937Z",{"id":129,"title":130,"url":131,"summary":132,"summary_zh":133,"content":8,"source_name":134,"source_url":131,"published_at":135,"category":12,"cover_url":8,"hotness":13,"is_selected":14,"score":136,"score_detail":137,"sources":142,"tags":144,"primary_topic":147,"secondary_topics":148,"classification":149,"search_phrases":168,"slug":171,"view_count":21,"doi":172,"paper":173,"created_at":206},5147,"FDUNet-LRSA: efficient frequency-adaptive segmentation for automated crop disease inspection","https:\u002F\u002Fdoi.org\u002F10.3389\u002Ffpls.2026.1963106","The transition toward industrialized and automated agriculture requires reliable machine vision for high-throughput crop health monitoring and field decision support. Automated sweetpotato disease inspection, however, remains challenging because field images exhibit substantial variations in leaf scale, appearance, illumination, occlusion, and background complexity. Conventional U-shaped networks rely primarily on fixed convolutional kernels and local feature aggregation, whereas Transformer-based alternatives often introduce considerable computational overhead. This article proposes FDUNet-LRSA, a frequency-adaptive segmentation network that integrates frequency-domain dynamic convolution, low-resolution self-attention, and boundary-aware refinement. Frequency-domain dynamic convolution reconstructs input-dependent spatial kernels to improve feature adaptation under changing imaging conditions. Low-resolution self-attention captures long-range contextual dependencies in a fixed compact feature space, while boundary-aware refinement preserves irregular leaf contours during mask reconstruction. Under the reported evaluation protocol, FDUNet-LRSA achieved a mean test foreground IoU of 0.94 and a Dice coefficient of 0.97, with 3.72 GFLOPs. Additional experiments indicate that task-oriented moderate augmentation improves generalization compared with stronger augmentation. The proposed network provides an efficient visual perception approach for automated crop inspection systems.","向工业化与自动化农业转型，需要可靠的机器视觉技术，以支持高通量作物健康监测与田间决策。然而，甘薯病害自动检测仍具挑战性，因为田间图像在叶片尺度、外观、光照、遮挡和背景复杂度方面存在显著变化。传统U形网络主要依赖固定卷积核和局部特征聚合，而基于Transformer的替代方案往往带来较大的计算开销。本文提出FDUNet-LRSA，一种频率自适应分割网络，融合了频域动态卷积、低分辨率自注意力与边界感知细化。频域动态卷积重建依赖输入的空间卷积核，以提升成像条件变化下的特征适应性。低分辨率自注意力在固定的紧凑特征空间中捕获长程上下文依赖，而边界感知细化在掩码重建过程中保留不规则叶片轮廓。在所报告的评估协议下，FDUNet-LRSA取得了0.94的平均测试前景IoU和0.97的Dice系数，计算量为3.72 GFLOPs。额外实验表明，与更强的增强相比，面向任务的中等增强可提升泛化能力。所提出的网络为自动化作物检测系统提供了一种高效的视觉感知方法。","Frontiers in Plant Science","2026-10-16T00:00:00Z",76,{"impact":19,"substance":138,"depth":139,"authority":76,"freshness":140,"relevant":21,"comment":141},21,17,9,"提出频域自适应分割网络，在甘薯病害检测上取得高精度且计算量低，方法有新意、结论可靠，属细分领域实质进展，值得进入每日精选。",[143],{"name":134,"url":131},[26,81,27,145,146],"甘薯","图像分割",{"code":30,"slug":31,"name":32,"path":33,"parent_code":34},[],{"status":37,"locked":14,"version":38,"method":39,"evidence":150},{"candidates":151,"agricultural_evidence":44},[152,158,164],{"code":30,"score":153,"supported":44,"title":154,"abstract":155,"tags":157},5,[],[156],"decision support",[81,146],{"code":159,"score":160,"supported":44,"title":161,"abstract":162,"tags":163},"A5",4,[],[145],[145],{"code":51,"score":21,"supported":14,"title":165,"abstract":166,"tags":167},[],[],[26],[169,170],"FDUNet-LRSA 甘薯病害","甘薯病害 图像分割","FDUNet-LRSA甘薯病害-5147","10.3389\u002Ffpls.2026.1963106",{"doi":172,"openalex_id":174,"authors":175,"venue":134,"cited_by_count":109,"oa_url":198,"card":199,"direction":205,"ingested_from":126},"W7222502318",[176,179,182,185,187,190,192,194,196],{"name":177,"orcid":178},"Ru Han","https:\u002F\u002Forcid.org\u002F0009-0000-5128-1314",{"name":180,"orcid":181},"Lei Shu","https:\u002F\u002Forcid.org\u002F0000-0002-6700-9347",{"name":183,"orcid":184},"Grzegorz Cielniak","https:\u002F\u002Forcid.org\u002F0000-0002-6299-8465",{"name":186,"orcid":8},"Yan Zhu",{"name":188,"orcid":189},"Jucheng Yang","https:\u002F\u002Forcid.org\u002F0000-0002-5046-2663",{"name":191,"orcid":8},"Fan Yang",{"name":193,"orcid":8},"Weihan Song",{"name":195,"orcid":8},"Qian Zhang",{"name":197,"orcid":8},"Kai Huang","https:\u002F\u002Fwww.frontiersin.org\u002Fjournals\u002Fplant-science\u002Farticles\u002F10.3389\u002Ffpls.2026.1963106\u002Fpdf",{"tldr":200,"method":201,"finding":202,"direction":203,"opportunity":204},"提出频域自适应分割网络FDUNet-LRSA，用于甘薯病害自动巡检。","频域动态卷积、低分辨率自注意力与边界感知细化，结合适度数据增强。","测试前景IoU达0.94、Dice达0.97，仅3.72 GFLOPs，轻量高效。","农业遥感与作物表型","可探索频域动态卷积在多种作物病害与复杂田间光照下的跨域泛化与边缘部署。","智慧农业 \u002F 农业物联网","2026-10-17T23:30:08.487857Z",{"id":208,"title":209,"url":210,"summary":211,"summary_zh":8,"content":8,"source_name":212,"source_url":210,"published_at":213,"category":12,"cover_url":8,"hotness":13,"is_selected":14,"score":214,"score_detail":215,"sources":219,"tags":221,"primary_topic":222,"secondary_topics":223,"classification":224,"search_phrases":237,"slug":240,"view_count":160,"doi":241,"paper":242,"created_at":251},1727,"Deep Learning-Based Crop Disease Detection Using EfficientNet-B3 for Smart Agriculture","https:\u002F\u002Fdoi.org\u002F10.21203\u002Frs.3.rs-10104203\u002Fv1","Deep Learning-Based Crop Disease Detection Using EfficientNet-B3 for Smart Agriculture。Research Square","Research Square","2026-09-04T00:00:00Z",56,{"impact":17,"substance":43,"depth":216,"authority":217,"freshness":59,"relevant":21,"comment":218},14,8,"基于EfficientNet-B3的深度学习作物病害检测研究，方法有一定新颖性，但缺乏具体数据规模与结论细节，权威性一般。",[220],{"name":212,"url":210},[26,81,27],{"code":30,"slug":31,"name":32,"path":33,"parent_code":34},[],{"status":37,"locked":14,"version":38,"method":39,"evidence":225},{"candidates":226,"agricultural_evidence":44},[227,232],{"code":51,"score":17,"supported":44,"title":228,"abstract":230,"tags":231},[229],"smart agriculture",[229],[26],{"code":30,"score":17,"supported":44,"title":233,"abstract":235,"tags":236},[234],"deep learning",[234],[81],[238,239],"农业人工智能 作物病害 智慧农业","农业人工智能 作物病害","农业人工智能作物病害智慧农业-1727","10.21203\u002Frs.3.rs-10104203\u002Fv1",{"doi":241,"openalex_id":243,"authors":244,"venue":212,"cited_by_count":109,"oa_url":210,"card":8,"direction":205,"ingested_from":126},"W7208777285",[245,247,249],{"name":246,"orcid":8},"Gajanan Ankatwar",{"name":248,"orcid":8},"Chitra Dhawale",{"name":250,"orcid":8},"Supriya Narad","2026-09-05T23:30:09.738110Z",{"id":253,"title":254,"url":255,"summary":256,"summary_zh":8,"content":8,"source_name":212,"source_url":255,"published_at":257,"category":12,"cover_url":8,"hotness":13,"is_selected":14,"score":258,"score_detail":259,"sources":261,"tags":263,"primary_topic":265,"secondary_topics":270,"classification":272,"search_phrases":288,"slug":290,"view_count":160,"doi":291,"paper":292,"created_at":305},1404,"Explainable Deep Learning-Based Potato Leaf Disease Detection and Severity Assessment for Smart Agriculture in Bangladesh","https:\u002F\u002Fdoi.org\u002F10.21203\u002Frs.3.rs-10879535\u002Fv1","Explainable Deep Learning-Based Potato Leaf Disease Detection and Severity Assessment for Smart Agriculture in Bangladesh。Research Square","2026-09-01T00:00:00Z",53,{"impact":217,"substance":43,"depth":43,"authority":217,"freshness":59,"relevant":21,"comment":260},"针对孟加拉国马铃薯病害的可解释深度学习检测研究，方法新颖但地域性强，影响有限。",[262],{"name":212,"url":255},[26,81,264,27],"马铃薯",{"code":159,"slug":266,"name":267,"path":268,"parent_code":269},"tubers-grains","薯类与杂粮","\u002Ftopic\u002Ffield-crops\u002Ftubers-grains","A",[271],{"code":30,"slug":31,"name":32,"path":33,"parent_code":34},{"status":37,"locked":14,"version":38,"method":39,"evidence":273},{"candidates":274,"agricultural_evidence":44},[275,280,284],{"code":159,"score":17,"supported":44,"title":276,"abstract":278,"tags":279},[277],"potato",[277],[264],{"code":51,"score":17,"supported":44,"title":281,"abstract":282,"tags":283},[229],[229],[26],{"code":30,"score":17,"supported":44,"title":285,"abstract":286,"tags":287},[234],[234],[81],[289,239],"农业人工智能 作物病害 智慧农业 马铃薯","农业人工智能作物病害智慧农业马铃薯-1404","10.21203\u002Frs.3.rs-10879535\u002Fv1",{"doi":291,"openalex_id":293,"authors":294,"venue":212,"cited_by_count":109,"oa_url":304,"card":8,"direction":205,"ingested_from":126},"W7204884709",[295,298,301],{"name":296,"orcid":297},"Kazi Abdul Mannan","https:\u002F\u002Forcid.org\u002F0000-0002-7123-132X",{"name":299,"orcid":300},"Sinha Nafiz Hasan","https:\u002F\u002Forcid.org\u002F0009-0007-8906-4586",{"name":302,"orcid":303},"Ahmed Shafin","https:\u002F\u002Forcid.org\u002F0009-0000-4162-6913","https:\u002F\u002Fwww.researchsquare.com\u002Farticle\u002Frs-10879535\u002Flatest.pdf","2026-09-02T23:30:10.900822Z",{"id":307,"title":308,"url":309,"summary":310,"summary_zh":311,"content":8,"source_name":71,"source_url":309,"published_at":72,"category":12,"cover_url":8,"hotness":13,"is_selected":14,"score":312,"score_detail":313,"sources":315,"tags":317,"primary_topic":321,"secondary_topics":327,"classification":329,"search_phrases":345,"slug":348,"view_count":109,"doi":349,"paper":350,"created_at":364},5592,"Evaluation of Concept-Rank Collapse Audits for Computer Vision Agriculture Disease Classifiers Under Low-Label Regimes: An Interpretability Capacity Bound for Smallholder Fungicide Selectivity","https:\u002F\u002Fdoi.org\u002F10.5281\u002Fzenodo.23545193","Concept-based interpretability methods for plant-disease classifiers are typically evaluated by whether their concept activations correlate with human-annotated labels, but this criterion is silent on a prior question: whether the classifier's internal representation retains enough concept-rank structure for any audit to be informative. the present analysis formalises that prior question as an interpretability capacity bound. A classifier is modelled as a concept-rank operator whose singular values govern how much concept-level signal survives the projection from pixel space to class logits, and a collapse audit is defined as the ratio of retained concept-rank energy to total concept energy. Under a low-label regime, where concept annotations are scarce and the audit must be estimated from few labelled examples, the bound predicts that audit reliability degrades as a function of the concept-rank spectrum and the label budget. The analysis is a theoretical and methodological contribution: it derives the bound, states the conditions under which a collapse audit is identifiable, and characterises the operating characteristics of the audit as a decision procedure under a declared variance structure (between-cluster SD = 6.0, residual SD = 4.5, intraclass correlation = 0.64). Power was estimated across 2000 design cells for cluster counts of 8, 16, 24 and 32 and effect sizes of 3, 5 and 8 percentage points; 52 clusters were required for 5 percentage points at 80% power. The bound reframes concept audits as capacity tests rather than fidelity tests, with consequences for how smallholder fungicide-selectivity decisions should be supported by interpretable disease classifiers.","面向植物病害分类器的基于概念的可解释性方法，通常以其概念激活是否与人工标注标签相关作为评价标准，但该标准对一个更先前的问题保持沉默：分类器的内部表征是否保留了足够的概念秩结构，从而使任何审计具有信息量。本分析将该先前问题形式化为一个可解释性容量上界。分类器被建模为概念秩算子，其奇异值支配着从像素空间到类别logits的投影中概念级信号得以保留的程度，并将坍缩审计定义为保留的概念秩能量与总概念能量之比。在低标签情形下，概念标注稀缺且审计必须由少量标注样本估计，该上界预测审计可靠性会随概念秩谱和标签预算的函数而退化。本分析是一项理论和方法论贡献：推导出该上界，陈述坍缩审计可识别的条件，并在声明的方差结构（簇间SD = 6.0，残差SD = 4.5，组内相关系数 = 0.64）下刻画审计作为决策程序的操作特征。在2000个设计单元中，针对簇数为8、16、24和32以及效应量为3、5和8个百分点的情形估计了功效；在80%功效下达到5个百分点需要52个簇。该上界将概念审计重新框定为容量检验而非保真度检验，并对可解释病害分类器应如何支持小农杀菌剂选择性决策具有启示意义。",69,{"impact":217,"substance":138,"depth":18,"authority":17,"freshness":13,"relevant":21,"comment":314},"理论方法论文，提出可解释性容量上界并给出审计功效分析，对智慧农业病害识别可信度有参考价值，但属小众方法学研究、缺乏田间实证，暂不宜进入每日精选。",[316],{"name":71,"url":309},[26,81,318,319,320],"小农户","植物病害识别","可解释性",{"code":322,"slug":323,"name":324,"path":325,"parent_code":326},"G2","farm-services","社会化服务与规模经营","\u002Ftopic\u002Frural-services\u002Ffarm-services","G",[328],{"code":30,"slug":31,"name":32,"path":33,"parent_code":34},{"status":37,"locked":14,"version":38,"method":39,"evidence":330},{"candidates":331,"agricultural_evidence":44},[332,337,341],{"code":322,"score":17,"supported":44,"title":333,"abstract":335,"tags":336},[334],"smallholder",[334],[318],{"code":30,"score":13,"supported":44,"title":338,"abstract":339,"tags":340},[93],[],[81,319],{"code":51,"score":21,"supported":14,"title":342,"abstract":343,"tags":344},[],[],[26],[346,347],"农业人工智能 植物病害识别 可解释性 智慧农业","农业人工智能 植物病害识别","农业人工智能植物病害识别可解释性智慧农业-5592","10.5281\u002Fzenodo.23545193",{"doi":349,"openalex_id":351,"authors":352,"venue":71,"cited_by_count":109,"oa_url":309,"card":359,"direction":124,"ingested_from":126},"W7223137323",[353,355,357],{"name":354,"orcid":8},"Akshay Ramgoolam",{"name":356,"orcid":8},"Priya Bundhoo Appadoo",{"name":358,"orcid":8},"Jean-Noël LEspérance",{"tldr":360,"method":361,"finding":362,"direction":124,"opportunity":363},"提出可解释性容量上界，评估低标签下概念秩崩溃审计的可靠性。","将分类器建模为概念秩算子，推导容量上界并做2000格功效模拟。","审计可靠性随概念秩谱与标签预算下降；5个百分点效应需52个聚类达80%功效。","可将容量上界用于小农户杀菌剂选择性决策的可解释模型审计与标签预算设计。","2026-10-22T11:21:10.425290Z",{"id":366,"title":367,"url":368,"summary":369,"summary_zh":370,"content":8,"source_name":71,"source_url":368,"published_at":72,"category":12,"cover_url":8,"hotness":13,"is_selected":14,"score":312,"score_detail":371,"sources":373,"tags":375,"primary_topic":379,"secondary_topics":385,"classification":387,"search_phrases":410,"slug":413,"view_count":109,"doi":414,"paper":415,"created_at":427},5591,"Hierarchical Conformal Risk Control for Artificial Intelligence Agriculture Irrigation Scheduling Under Unmodeled Infiltration: A Computer Safety-Guarantee Evaluation","https:\u002F\u002Fdoi.org\u002F10.5281\u002Fzenodo.23543665","Irrigation scheduling algorithms that rely on soil-moisture forecasts are vulnerable to unmodeled infiltration, the process by which water bypasses the root zone through preferential pathways that resistance-based soil models cannot represent. The framework is evaluated through a replication-based design study that characterises the operating characteristics of the before-after-control-impact estimator used to audit scheduler performance, under a declared variance structure with between-cluster standard deviation 6.0, residual standard deviation 4.5 and intraclass correlation 0.64. Power was estimated across 2000 design cells spanning cluster counts of 8, 16, 24 and 32 and effect sizes of 3, 5 and 8 percentage points. The analysis indicates that 54 clusters are required to detect a 5-percentage-point effect at 80% power, and that the minimum detectable effect declines steeply as cluster count rises from 8 to 32. The contribution is a formal safety guarantee for irrigation decision support that remains valid when the infiltration model is wrong, together with an audit design that makes the guarantee empirically checkable.","依赖土壤湿度预报的灌溉调度算法容易受到未建模入渗的影响，即水分通过优先路径绕过根区的过程，而基于阻力的土壤模型无法表征这一过程。该框架通过一项基于重复的设计研究进行评估，该研究刻画了用于审计调度器性能的前后对照影响估计量的运行特性，其所依据的声明方差结构为：组间标准差6.0、残差标准差4.5、组内相关系数0.64。功效估计覆盖2000个设计单元，涵盖聚类数8、16、24和32以及效应量3、5和8个百分点。分析表明，在80%功效下检测5个百分点效应需要54个聚类，且最小可检测效应随聚类数从8增至32而急剧下降。本文的贡献在于为灌溉决策支持提供了一种正式的安全保证，该保证在入渗模型错误时仍然有效，同时提出了一种使该保证可经实证检验的审计设计。",{"impact":217,"substance":75,"depth":18,"authority":17,"freshness":140,"relevant":21,"comment":372},"该研究提出在入渗模型失配条件下仍有效的灌溉决策安全保证框架，并给出可审计的试验设计，方法新颖、结论可靠，但属细分方法学研究，产业影响有限，适合作为专业精选而非头条。",[374],{"name":71,"url":368},[26,81,376,377,378],"灌溉决策","安全保证","试验设计",{"code":380,"slug":381,"name":382,"path":383,"parent_code":384},"H3","irrigation","水资源与灌溉","\u002Ftopic\u002Fenvironment\u002Firrigation","H",[386],{"code":30,"slug":31,"name":32,"path":33,"parent_code":34},{"status":37,"locked":14,"version":38,"method":39,"evidence":388},{"candidates":389,"agricultural_evidence":44},[390,395,399,406],{"code":380,"score":43,"supported":44,"title":391,"abstract":392,"tags":394},[381],[393,381],"灌溉",[376],{"code":30,"score":17,"supported":44,"title":396,"abstract":397,"tags":398},[92],[156],[81],{"code":400,"score":63,"supported":44,"title":401,"abstract":402,"tags":405},"H2",[],[403,404],"土壤","soil",[],{"code":51,"score":21,"supported":14,"title":407,"abstract":408,"tags":409},[],[],[26],[411,412],"灌溉调度 安全保证","农业人工智能 安全保证 智慧农业 灌溉决策","灌溉调度安全保证-5591","10.5281\u002Fzenodo.23543665",{"doi":414,"openalex_id":416,"authors":417,"venue":71,"cited_by_count":109,"oa_url":368,"card":422,"direction":124,"ingested_from":126},"W7223131205",[418,420],{"name":419,"orcid":8},"Nassima Benkhelifa",{"name":421,"orcid":8},"Yacine Aït Ahmed",{"tldr":423,"method":424,"finding":425,"direction":124,"opportunity":426},"提出分层保形风险控制，为未建模入渗下的灌溉调度提供安全保证。","分层保形风险控制与基于重复设计的BACI审计，2000个设计单元功效分析。","检测5个百分点效应需54个集群，最小可检测效应随集群数增加陡降。","可将保形风险控制扩展到其他未建模扰动下的农业决策安全保证与审计设计。","2026-10-22T11:21:10.283561Z"]