Early and accurate detection of pigeon pea leaf diseases is essential for improving crop productivity and ensuring food security, particularly under real-field agricultural conditions. This paper introduces a shallow and computationally off-the-shelf deep learning system to detect the presence of pigeon pea leaf disease with great accuracy and in real-time on resource-limited cameras. DSLR and smartphone cameras were used to make up a custom high-resolution dataset under natural field conditions, including healthy leaves and major diseases, such as Fusarium wilt, leaf spot, and powdery mildew. All the images were downsampled to 224 × 224 pixels and processed with a Gaussian smoothing filter to remove noise and a Canny edge detector to improve structural features. Disease regions were accurately isolated using a Skill Optimization Algorithm (SOA)-driven segmentation strategy that dynamically optimized threshold levels, morphological kernel sizes, and lesion area constraints to handle background clutter and illumination variations. A pretrained EfficientNet-B0 model was used to extract deep semantic features, which consisted of compact 1280-dimensional feature vectors. A novel FMDDCN approach was used to classify these features through exploiting the sensitivity to subtle disease patterns by relying on differential feature modeling and multi-layer fusion of features. The model was fitted on stochastic gradient descent with a learning rate of 1 x 10-3 and a batch size of 32, and assessed on a 60/20/20 train validation test split with 5-fold cross-validation. The results of the experiment show consistent convergence with low overfitting. The proposed framework was found to produce a classification accuracy of 94.5%, precision of 91.0%, recall of 85.5% and Matthews Correlation Coefficient of 88.5% when it was used with four optimized features. In comparison, it is demonstrated that FMDDCN performs better than traditional machine learning and deep learning models, with its F1-score of 0.965 and the overall accuracy of 0.965. The suitability of the real-time edge deployment is verified, as confirmed by the use of computational analysis to reduce inference latency and memory consumption.
Why it matches plant phenotyping methods画像から植物葉の病徴・病害状態を推定するリアルタイム手法の開発と技術評価が中心であり、植物フェノタイピング手法に該当する。
abstractThis paper introduces a shallow and computationally off-the-shelf deep learning system to detect the presence of pigeon pea leaf disease with great accuracy and in real-time on resource-limited cameras.
Common beanPigeon peaLaboratory / benchtopMultispectral / hyperspectralSeed / grainClassification
Reliable seed accession identification underpins germplasm conservation, traceability and breeding; however, conventional assays remain destructive, labour-intensive and difficult to scale. Here, visible-near-infrared-shortwave infrared (VIS-NIR-SWIR) hyperspectral imaging (HSI; 449.54-2399.17 nm; 563 bands) was used to classify 32 grain-legume accessions ( n = 3200 seeds; 100 seeds per accession), comprising 30 common bean ( Phaseolus vulgaris L.) landraces plus two outgroup legumes ( Vigna angularis (Willd.) Ohwi & Ohashi and Cajanus cajan (L.) Huth). Each seed was represented by one ROI-averaged spectrum obtained from mean representative pixels within a standardised 10 × 10 pixel window at the centre of each seed. A fixed stratified 70:30 seed-level training:test partition was used, with 70 seeds per accession ( n = 2240) reserved for fully independent training and 30 seeds per accession ( n = 960) reserved as a fully independent test set. Principal component analysis (PCA) captured 97.42% of the spectral variance in the first three components (PC1 = 63.34%, PC2 = 23.78%, and PC3 = 10.31%). One-versus-rest wavelength association mapping revealed a maximum R 2 of 0.775 at 461.37 nm, and ReliefF concentrated the strongest reduced-band signal within 449.54-456.30 nm and 577.02-597.54 nm. In the original ReliefF-selected 16-band benchmark, the subspace discriminant reached 68.25% macro-F1 and 68.54% balanced accuracy; after edge-band trimming, the alternative 16-band configuration decreased to 60.67% and 60.94%, respectively. With respect to the full-spectrum sensitivity benchmark, linear discriminant analysis achieved 96.35% balanced accuracy, followed by linear SVM (94.17%). Deep learning trained directly on the full 563-band spectra reached 84.90% test accuracy, 84.47% macro-F1, 86.27% precision and 84.90% recall, with MLP_Wide outperforming the convolutional, recurrent and attention-based alternatives. Overall, under controlled laboratory conditions, this benchmark shows that accession discrimination is driven mainly by visible-domain contrasts in the most compact representations, whereas the full spectral context remains important for the most confusable accessions and for cautious future sensor design. The reduced-band findings should therefore be interpreted as exploratory guidance for sensor design rather than as a validated deployment-ready specification.
Why it matches plant phenotyping methods豆類種子の識別・分類を目的に、ハイパースペクトル画像取得、波長選択、機械学習・深層学習をベンチマークしており、種子形質の計測・抽出手法が中心である。
abstractOverall, under controlled laboratory conditions, this benchmark shows that accession discrimination is driven mainly by visible-domain contrasts
Phytophthora stem blight (PSB), caused by Phytophthora cajani, is a destructive disease of pigeonpea (Cajanus cajan L.) that can lead to complete crop loss under favorable conditions. Effective resistance breeding is constrained by the lack of reliable and reproducible screening methods. The present study aimed to standardize and compare artificial inoculation techniques for consistent induction of Phytophthora blight under controlled conditions. Ten inoculation methods targeting different infection pathways were evaluated in greenhouse pot experiments over two consecutive seasons (2023-24 and 2024-25) using susceptible (UPAS-120, ICP-7119, ICP-2376) and moderately resistant (KPBR-80-2-1, IPAC-79, IPAB-7-2-1-7) pigeonpea cultivars. Disease incidence was recorded to assess the efficiency and reproducibility of each technique. All methods successfully established infection; however, disease severity varied significantly among techniques and genotypes. The leaf-lamina inoculation method consistently produced the highest and most uniform disease incidence, recording 97.7% and 96.6% in the susceptible cultivar UPAS-120 and 83.3% and 86.6% in the moderately resistant cultivar KPBR-80-2-1 across the two seasons. Node inoculation emerged as the second most reliable method. The study identifies most efficient and quick, reproducible inoculation techniques that enable rapid and high-throughput resistance screening, providing a robust methodological framework to support pigeonpea breeding and Phytophthora blight management.
Why it matches plant phenotyping methods植物病害抵抗性の表現型（発病 incidence/severity）を取得する人工接種法を標準化・比較評価しており、フェノタイピング手法自体が中心である。
abstractThe present study aimed to standardize and compare artificial inoculation techniques for consistent induction of Phytophthora blight under controlled conditions.
Pigeon pea ( Cajanus cajan [L.] Millsp.) remains an underutilized legume in most African countries despite its potential to promote climate-resilient farming, diversify food sources, and enhance nutrition. Limited understanding of its indigenous diversity and farmer trait preferences hampers wider adoption, especially in the West African region. Between February and May 2025, a germplasm exploration was conducted across 18 Nigerian states, supplemented by accessions from the International Institute of Tropical Agriculture (IITA) genebank, Ghana, the Republic of Benin, and The Gambia, totaling 273 accessions. Ethnobotanical surveys documented farmer preferences, cultural uses, and local nomenclature, while seed morphometric traits were assessed using Videometerlab4 multispectral imaging. Farmer surveys revealed that cooking time (58.3%), commercial value (27.0%), and maturity cycle (14.7%) were key preferred traits. Gender and age influenced preferences; women and older farmers prioritized cooking time, whereas men and younger farmers emphasized the maturity cycle. Vernacular names (e.g., Otili, Fiofio, Waken Gwari ) reflected deep cultural ties and cross-border exchange in Ogun State and the Republic of Benin, highlighting transboundary diversity. Morphometric analysis showed moderate variation in seed size, shape, and color. Seed area (14.2–46.0 mm2), compactness (0.590–0.998), and eccentricity (0–0.808) distinguished rounded from elongated seeds, while CIELab_A values (− 0.04 to 29.98) captured color differences. The first two PCA axes explained 67.1% of the total variation, and cluster analysis grouped accessions into four morphotypes. By combining genetic, morphometric, and farmer preference data, this study offers a strong basis for conserving and developing climate-resilient, fast-cooking, and market-preferred cultivars for sub-Saharan Africa.
Why it matches plant phenotyping methodsVideometerlab4マルチスペクトル画像を用いた種子形態形質の取得と解析が研究の主要な構成要素であり、複数の形態・色指標と形態型分類を実施しているため、フェノタイピング手法の実質的な適用に該当する。
titleGermplasm exploration and digital phenotyping reveal indigenous diversity and farmer preferences in pigeon pea (Cajanus cajan (L.) Millsp.) for climate-smart breeding
Plant phenotyping relevance match · UnverifiedOpenAlex · Europe PMC · checked 6 Sept 2026
Abstract Pigeon pea ( Cajanus cajan [L.] Millsp.) remains an underutilized legume in most African countries despite its potential for climate-resilient farming systems, food diversification, and nutritional value. Limited knowledge of its indigenous diversity and farmer trait preference constrains wider adoption, particularly in the West African sub-region. Between February and June 2025, a germplasm exploration was conducted across 18 Nigerian states, complemented by accessions from the International Institute of Tropical Agriculture (IITA) genebank, Ghana, the Republic of Benin, and the Gambia, bringing the total to 273 accessions. Ethnobotanical surveys captured farmer preferences, cultural uses, and local nomenclature while seed morphometric traits were assessed using Videometerlab4 multispectral imaging. Farmer surveys revealed cooking time (58.3%), commercial value (27.0%), and maturity cycle (14.7%) as preferred varietal traits. Gender and age differences were evident; women and older farmers prioritized cooking time, while men and youth emphasized the maturity cycle as a preferred trait. Vernacular names (e.g., Otili , Fiofio , Waken Gwari ) highlighted deep cultural integration and cross-border exchange in Ogun State and the Republic of Benin, indicating transboundary diversity. Morphometric analyses revealed moderate variability in seed size, shape, and pigmentation. Seed area (14.2–46.0mm 2 ), Compactness (0.590–0.998), and eccentricity (0–0.808) differentiated rounded from elongated seeds, while CIELab_A values (–0.04–29.98) captured pigmentation differences. The first two PCA axes explained 67.1% of total variation, and cluster analysis grouped accessions into four morphotypes. By integrating genetic and morphometric information, as well as farmer varietal preference insights, this study provides a robust foundation for the conservation and development of climate-resilient, fast-cooking, and market-preferred varieties for sub-Saharan Africa.
Why it matches plant phenotyping methodsVideometerlab4によるマルチスペクトル画像から種子形態・色素形質を抽出し、PCAとクラスタリングで遺伝資源を分類するデジタル表現型解析が、研究の主要な構成要素である。
abstractseed morphometric traits were assessed using Videometerlab4 multispectral imaging
ABSTRACT Plant diseases are considered one of the most serious problems in world agricultural production. Regular monitoring and detection are essential to control plant diseases, and effective management methods are used to prevent disease spread and lower pesticide costs. Smart agriculture techniques are one of the key solutions in plant disease prediction and improving crop productivity. Even though various papers have been published on the model for plant disease prediction based on smart agriculture, there is still a lack of an overall systematic model. The proposed approach has been developed to overcome the challenges faced by the existing method. This presented approach uses deep learning and meta‐heuristic techniques to detect and classify crop diseases, providing an accurate and efficient solution for farmers to improve crop yield. The process begins with collecting crop disease images from the Kaggle database. Initially, noise removal and contrast enhancement are performed using a Gaussian Amended Wiener Filter (GAWF). Next, the Modified Residual U‐Net (MRU‐Net) model extracts significant disease regions from the images. Effective features are collected from these segments using a convolutional neural network (CNN) and an improved vision transformer model (IViT). Finally, classification is performed with a stacking ensemble model that incorporates XGBoost (XGB), Gradient Boosting (GB) and AdaBoost‐Decision Tree (AdB‐DT). The proposed model achieved an accuracy of 99.74% on the PlantVillage dataset, 99.51% on the PlantDoc dataset and 99.57% on the Pigeonpea Leaf Disease dataset, demonstrating its robustness and generalizability across both curated and real‐world agricultural image conditions. Also, the proposed approach provided insights into disease identification by utilising Grad‐CAM to provide visual explanations.
Why it matches plant phenotyping methods植物画像から病害領域を抽出・分類する深層学習パイプラインが研究の中心であり、植物の病害状態を直接推定して複数データセットで性能検証している。
abstractThe process begins with collecting crop disease images from the Kaggle database. Initially, noise removal and contrast enhancement are performed using a Gaussian Amended Wiener Filter (GAWF). Next, the Modified Residual U‐Net (MRU‐Net) model extracts significant disease regions from the images.
Diseases and pests in plants/crops are major causes of significant agricultural losses with economic, social and ecological impacts. Therefore, there is a need for early identification of plant diseases and pests through automated systems. Recently, machine learning-based methods have become popular in solving agricultural problems such as plant diseases faced by technically-noob farmers. This work proposes a novel method based on stacking ensemble machine learning to detect plant diseases in Uradbean precisely. Two classifiers: support vector machine (SVM), random forest (RF) are trained on a dataset consists of Uradbean infected and healthy leaf images. These classifiers are stacked with logistic regression (LR) classifier. In the diverse ensemble, LR classifier is used as a meta-learner which enhanced the precision of the disease classification. The fuzzy C-Means clustering with particle swarm optimization is used for image segmentation. Haralick, Hu Moments and color histogram methods are used in feature extraction. During the tests, the proposed model is also compared with pre-trained networks: DenseNet-201, ResNet-50, and VGG19. It achieved an impressive classification accuracy of 96.82 % which is higher than the individual classifiers and pre-trained networks. To validate model performance, it is evaluated on a benchmark public dataset consists of Apple leaf images and achieved 98.30% accuracy. It is observed that ensemble method reflects an advantage over individual models in increasing the classification rates and reducing the computational overhead in comparison to pre-trained networks which struggle due to the issues such as irrelevant features, generation of pertinent characteristics, and noise
Why it matches plant phenotyping methods葉画像から植物病害状態を推定する画像解析・機械学習手法の開発とベンチマーク検証が中心であり、植物フェノタイピング手法に該当する。
abstractThis work proposes a novel method based on stacking ensemble machine learning to detect plant diseases in Uradbean precisely.
Pigeonpea (Cajanus cajan), a legume of nutritional significance, is highly prone to wilt disease caused by fungal pathogen, Fusarium udum, that leads to 15–30 % of crop mortality in India. While early detection of wilts in legume is crucial for remedial measures, it has been poorly addressed till date using traditional field based manual methods. The present study aimed to design an integrated two-step wilt detection methodology, and develop a disease-specific spectral index for Cajanus cajan exploiting spectral enrichment of ASI-PRISMA hyperspectral dataset. Initially, Modified Red Edge Normalized Difference Vegetation Index, Normalized Difference Nitrogen Index, and Photochemical Reflectance Index were combined for generation of relative agricultural stress map and in parallel, Minimum Noise Fraction transformation and Pixel Purity Index (PPI) based endmember maps/spectra were generated. Integration of high agricultural stress areas/pixels with PPI endmembers successfully established the desired spectrum for the diseased Cajanus cajan plants. Subsequently, the novel two-step methodology was validated through ground truthing. In addition, a plant (C. cajan)-specific normalised difference disease/stress index was developed for rapid assessment of C. cajan health status, after exhaustive search for band combinations and separability analysis. To assess the robustness of the proposed two-step methodology and spectral index for disease detection in Cajanus cajan, another site was investigated. A total of seven DLR DESIS and EnMAP, and ASI-PRISMA hyperspectral images were exploited using the proposed methodology for wilt detection in C. cajan. It was established from the field experiments that hyperspectral imaging could efficiently detect the wilted C. cajan plants in the area. In conclusion, using spaceborne hyperspectral images, developed disease spectral index values of ≤0.55 and agricultural stress values ≥ 3 could jointly detect the wilt at an early stage in C. cajan. When compared with commonly used multispectral satellite imageries, the developed methodology for hyperspectral imagery based signature analysis could efficiently detect the diseased Cajanus cajan plants at least 23 weeks in advance. This is the first report on employing satellite hyperspectral imagery for the detection of the wilt in C. cajan. The field deployment of hyperspectral imaging based precise foreknowledge regarding the wilt in legumes would help the stakeholders to make more informed decisions for quick mitigation.
Why it matches plant phenotyping methods衛星ハイパースペクトル画像から植物の萎凋・健康状態を推定する疾病特異的スペクトル指標と二段階検出法を開発し、地上真値および別地点で検証しており、植物表現型の取得・推定手法が中心である。
abstractThe present study aimed to design an integrated two-step wilt detection methodology, and develop a disease-specific spectral index for Cajanus cajan exploiting spectral enrichment of ASI-PRISMA hyperspectral dataset.
The escalating incidence of plant diseases presents considerable obstacles to the agricultural domain, resulting in substantial reductions in crop yield and posing a threat to food security. To address the pressing concern of Black Gram Plant Leaf Diseases (BPLD), this research endeavors to tackle disease classification through the application of a deep learning methodology. The approach leverages a comprehensive dataset that encompasses Anthracnose, Leaf Crinkle, Powdery Mildew, and Yellow Mosaic diseases, all of which affect the black gram crop. By employing this advanced technique, we aim to contribute valuable insights to combat BPLD effectively. Our research applies deep learning models, including Darknet-53, ResNet-101, GoogLeNet, and EfficientNet-B0, to classify plant diseases. Darknet-53 achieved 98.51% accuracy, followed by ResNet-101 (97.51%), GoogLeNet (96.52%), and EfficientNet-B0 (77.61%). These findings demonstrate the potential of deep learning for accurate disease identification, benefiting agriculture. The study provides a comparative analysis of deep learning models for Black Gram Plant Leaf Disease (BPLD) classification, revealing Darknet-53 and ResNet-101 as superior performers. Implementing these models in real-world agricultural scenarios holds promise for early disease detection and intervention, reducing potential crop losses. The high accuracy achieved signifies significant progress in automating disease recognition, benefiting the agricultural sector.
Why it matches plant phenotyping methods黒グラム葉の病徴を画像から深層学習で分類する手法を比較・評価しており、植物の病害状態推定が研究の中心であるため。
abstractthis research endeavors to tackle disease classification through the application of a deep learning methodology
The B value is required to quantify the nitrogen derived from the atmosphere (%Ndfa) in the Rhizobium-legume symbiosis using the 15N natural abundance method. When the B value of a particular specie is not known, one possibility Is to use as a proxy the B value of a specie from the same genus, but this can cause the estimate of %Ndfa to be inaccurate. In this work, we compared two methodologies for determining the B value of Crotalaria juncea, C. spectabilis, C. ochroleuca and Cajanus cajan, using soil as the substrate. One method involvedgrowing plants in soil and averaging the lowest {delta}15N values of plant shoots (B-minimum), while the other consisted in adding sucrose to soil to immobilize the mineral nitrogen (N-immobilized), and then averaging the shoot {delta}15N values of all plants. Results showed that B values of C. cajan and C. ochroleuca obtained using the N-immobilized method were up to 1{per thousand} lower than those reported in the literature for these species. Therefore, we propose that, at least in these species, B values determined with the N-immobilized method should be used to estimate the%Ndfa.
Why it matches plant phenotyping methodsマメ科植物の窒素固定量推定に用いるB値について、植物体のδ15N測定を基盤とする2手法を比較・検証しており、植物の生理状態の定量法が研究の中心である。
abstractIn this work, we compared two methodologies for determining the B value of Crotalaria juncea, C. spectabilis, C. ochroleuca and Cajanus cajan, using soil as the substrate.
Pigeonpea is an important legume cultivated in India, owing to its high nutritive and protein value. Various eco-friendly and green strategies have been proposed by researchers for enhancing the productivity of this pulse crop. In the present study, different bioformulations were prepared from plant growth promoting (PGP) bacterial strains and coated on pigeonpea seeds for monitoring growth enhancement of plantlets under in vitro conditions. For performing initial in vitro experiments with efficient bioinoculants, the conventional methods employ either tubes or Petri dishes. Despite delivering satisfactory results under laboratory conditions these methods suffer from inherent limitations of being cumbersome, difficult, and error-prone when multiple samples are handled simultaneously. The present study aimed to develop a high-throughput method for recording plant parameters of more than ninety seeds simultaneously, thus facilitating the process. The study opens up avenues of faster assessment to answer a range of scientific questions related to plant-microbe interactions and beyond.
Why it matches plant phenotyping methods植物パラメータを多数同時に記録する高スループット測定法の開発が研究の中心であり、単なるバイオ製剤評価ではない。
abstractThe present study aimed to develop a high-throughput method for recording plant parameters of more than ninety seeds simultaneously, thus facilitating the process.
Warm-season legumes have been receiving increased attention as forage resources in the southern United States and other countries. However, the near infrared spectroscopy (NIRS) technique has not been widely explored for predicting the forage quality of many of these legumes. The objective of this research was to assess the performance of NIRS in predicting the forage quality parameters of five warm-season legumes-guar ( Cyamopsis tetragonoloba ), tepary bean ( Phaseolus acutifolius ), pigeon pea ( Cajanus cajan ), soybean ( Glycine max ), and mothbean ( Vigna aconitifolia )-using three machine learning techniques: partial least square (PLS), support vector machine (SVM), and Gaussian processes (GP). Additionally, the efficacy of global models in predicting forage quality was investigated. A set of 70 forage samples was used to develop species-based models for concentrations of crude protein (CP), acid detergent fiber (ADF), neutral detergent fiber (NDF), and in vitro true digestibility (IVTD) of guar and tepary bean forages, and CP and IVTD in pigeon pea and soybean. All species-based models were tested through 10-fold cross-validations, followed by external validations using 20 samples of each species. The global models for CP and IVTD of warm-season legumes were developed using a set of 150 random samples, including 30 samples for each of the five species. The global models were tested through 10-fold cross-validation, and external validation using five individual sets of 20 samples each for different legume species. Among techniques, PLS consistently performed best at calibrating (R 2 c = 0.94-0.98) all forage quality parameters in both species-based and global models. The SVM provided the most accurate predictions for guar and soybean crops, and global models, and both SVM and PLS performed better for tepary bean and pigeon pea forages. The global modeling approach that developed a single model for all five crops yielded sufficient accuracy (R 2 cv /R 2 v = 0.92-0.99) in predicting CP of the different legumes. However, the accuracy of predictions of in vitro true digestibility (IVTD) for the different legumes was variable (R 2 cv /R 2 v = 0.42-0.98). Machine learning algorithms like SVM could help develop robust NIRS-based models for predicting forage quality with a relatively small number of samples, and thus needs further attention in different NIRS based applications.
Why it matches plant phenotyping methodsNIRSと機械学習を用いてマメ科飼料の品質形質を推定するモデルを開発し、交差検証と外部検証で性能評価しており、形質取得法が研究の中心である。
abstractThe objective of this research was to assess the performance of NIRS in predicting the forage quality parameters of five warm-season legumes