Smart agricultural technology最新文献

{"title":"Machine learning-based cotton yield forecasting under climate change for precision agriculture","authors":"Muhammad Umair Shahzad ,&nbsp;Sana Tahir ,&nbsp;Javed Rashid ,&nbsp;Osama A. Khashan ,&nbsp;Rashid Ahmad ,&nbsp;Sheikh Mansoor ,&nbsp;Anwar Ghani","doi":"10.1016/j.atech.2025.101117","DOIUrl":"10.1016/j.atech.2025.101117","url":null,"abstract":"<div><div>The escalating threat of climate change presents a significant challenge to modern agriculture, with serious consequences for global food security. The impact of changing climate variables on crop productivity, particularly for key agricultural commodities, raises concerns about future yields. This study examines the potential effects of climate change on cotton production by integrating historical climate data, Global Climate Models (GCMs, CMIP3) projections, and cotton yield data. This study employs a diverse range of machine learning (ML) methods, including multiple regression, k-nearest neighbors (KNN), boosted tree algorithms, and various types of artificial neural networks (ANNs), to investigate the intricate relationship between climate factors and cotton yields. The models are developed and tested using data on climate and crop yields collected from three regions in Punjab, Pakistan, spanning the years 1991 to 2020. To estimate future yield outcomes, climate projections from General Circulation Models (GCMs) are downscaled under the SRA1B, A2, and B1 carbon emission scenarios, enabling forecasts extending to the year 2050. Results show that rainfall has a negligible impact on cotton yield (R = 0.0002), whereas maximum temperature (R = -0.183) is identified as the primary climatic factor influencing yield, followed by minimum temperature (R = 0.248). Among the models, the generalized feedforward (GFF) demonstrated the best performance (R = 0.960, MSE = 0.110, NMSE = 0.187, MAE = 0.269), outperforming probabilistic neural network (PNN), KNN, multilayer perceptron (MLP), and boosted trees. In contrast, linear regression (LR) and multiple regression models performed less effectively. The reliability of GFF and KNN in providing yield estimates (<span><math><msup><mrow><mi>R</mi></mrow><mrow><mn>2</mn></mrow></msup></math></span> = 0.892, 0.861) supports their potential for accurate predictions. The study forecasts a 4.5% decline in cotton yield by 2050 compared to the highest recorded yield for the region, highlighting the impact of climate change on cotton production and its potential threat to food security. Nevertheless, the adaptive capabilities of the ANN (GFF) models across various climate scenarios present promising tools for integrating ML into climate-resilient agricultural practices, contributing to sustainable agrarian security and mitigating the adverse effects of climate change on food supply.</div></div>","PeriodicalId":74813,"journal":{"name":"Smart agricultural technology","volume":"12 ","pages":"Article 101117"},"PeriodicalIF":6.3,"publicationDate":"2025-06-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144519208","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}

{"title":"TD-CFD-DPM Coupled method for multi-objective optimization of collision pollination parameters in hybrid rice seed production","authors":"Te Xi ,&nbsp;Rongkai Shi ,&nbsp;Huaiqu Feng ,&nbsp;Bo Chen ,&nbsp;Nian Li ,&nbsp;Yongwei Wang ,&nbsp;Jun Wang","doi":"10.1016/j.atech.2025.101130","DOIUrl":"10.1016/j.atech.2025.101130","url":null,"abstract":"<div><div>Elucidation of the pollen dispersal behavior during the production of hybrid rice seeds is imperative for the optimization of the pollination process and the promotion of mechanized pollination. This paper presents a multi-objective optimization method combining the TD-CFD-DPM (Transient Dynamics - Computational Fluid Dynamics - Discrete Phase Model) method and genetic algorithm for optimizing collisional pollination parameters for large-scale seed production of hybrid rice. The construction of a rice-air-pollen multiphase coupling model was undertaken to simulate the pollen diffusion and deposition process. This model was then combined with a response surface experimental design to construct an objective function, which was used to visualize the pollen movement trajectory and deposition distribution. The genetic algorithm was further utilized to optimize the pollination operation parameters. The feasibility of the model and the optimized parameters was verified by means of collaborative collision module experiments. The results demonstrated that the optimized parameter combinations exhibited satisfactory performance with regard to pollen dispersal distance and distribution uniformity. The pollen dispersal distances were all greater than 1.3 m The uniformity of pollen distribution was high, and its coefficient of variation was maintained below 80 %. The mean discrepancy between the calculated and experimental values of the optimized parameter combinations was found to be &lt;6 %. This study offers both theoretical underpinnings and practical directives to advance the mechanized pollination theory and facilitate the mechanized pollination of hybrid rice for large-scale seed production.</div></div>","PeriodicalId":74813,"journal":{"name":"Smart agricultural technology","volume":"12 ","pages":"Article 101130"},"PeriodicalIF":6.3,"publicationDate":"2025-06-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144502518","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}

{"title":"PB-STR: A spatiotemporal transformer network for multi-behavior recognition of pigs","authors":"Yufan Hu ,&nbsp;Xiaobo Wang ,&nbsp;Rui Mao ,&nbsp;Yusen Guo ,&nbsp;Xianyao Zhu ,&nbsp;Meili Wang","doi":"10.1016/j.atech.2025.101131","DOIUrl":"10.1016/j.atech.2025.101131","url":null,"abstract":"<div><div>Pig behavior is a reliable indicator of health status, accurate recognition is vital for effective health surveillance and management. This study proposes PB-STR, a behavior recognition model based on the integration of video spatiotemporal feature fusion. The model addresses challenges in recognizing multiple behaviors within a single frame and handling dynamically changing behaviors. It develops a Time Series Prediction Module (UnetTSF) and a Context Anchor Attention (CAA) module, enhancing the PB-STR framework's ability to capture feature evolution over time and fully utilize contextual information. To enhance the model's proficiency in detecting and recognizing behaviors within overlapping regions, the detection head employs Minimum Points Distance Intersection over Union (MPDIoU) as its bounding box loss function, improving adaptability to variations in pig positions. The PB-STR model was evaluated on a proprietary dataset of 294 videos covering seven pig behaviors. With a mean Average Precision of 94.2 %, recall of 90.8 %, and precision of 87.5 %, the PB-STR model can concurrently recognize five dynamic and two static behaviors in pigs. By outperforming models such as DETR, DAB-DETR, Deformable DETR, CenterNet, and DINO, the proposed approach not only enhances detection accuracy but also serves as a technological foundation for intelligent, welfare-oriented pig farming, facilitating in the sector's modernization.</div></div>","PeriodicalId":74813,"journal":{"name":"Smart agricultural technology","volume":"12 ","pages":"Article 101131"},"PeriodicalIF":6.3,"publicationDate":"2025-06-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144502517","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}

{"title":"Comparison among grazing animal behavior classification algorithms for use with open-source wearable sensors","authors":"B.R. dos Reis ,&nbsp;S. Sujani ,&nbsp;D.R. Fuka ,&nbsp;Z.M. Easton ,&nbsp;R.R. White","doi":"10.1016/j.atech.2025.101133","DOIUrl":"10.1016/j.atech.2025.101133","url":null,"abstract":"<div><div>Behavioral monitoring for pasture-based production has the potential to improve the efficiency of livestock operations without increasing labor costs. The application of technologies for remotely monitoring animal behavior has expanded rapidly in the last decade, especially in confinement operations. However, automated systems for extensive operations are limited by the interrelated challenges of power use and data transmission. The objective of this study was to explore behavioral classification techniques suitable for using an open-source wearable sensor system deployed on extensively grazed cows. Behavior classification analyses leveraged simple approaches (analysis of variance and logistic regression), as well as more complex machine learning algorithms (support vector machine (SVM) and random forest (RF)) to better understand the trade-offs between classification approach complexity and accuracy. Behavioral observations were conducted by two independent observers at the field. Algorithms were used to classify four behaviors: grazing, lying, standing, and walking using data aggregated across either 1-second or 1-minute intervals. Algorithms were also compared under situations assuming continuous monitoring compared with periodic snapshots of data representing a scenario where the sensor was only activated every 3 or 5 s. Grazing was the most accurately (93 %) classified behavior followed by laying (92 %) when a 1-minute timestep was used to train the RF model. At both timesteps, SVM and RF were capable of distinguishing among behaviors with improved accuracy compared with simplistic approaches. Similar accuracies were found when evaluating the RF model on the 3 and 5-second iteration, indicating power saving may be achieved by periodic, rather than continuous sampling. As microprocessors continue to advance in terms of their capacity to execute machine learning algorithms, these approaches may help improve the usability of inertial measurement unit sensors for behavioral monitoring in extensive production systems.</div></div>","PeriodicalId":74813,"journal":{"name":"Smart agricultural technology","volume":"12 ","pages":"Article 101133"},"PeriodicalIF":6.3,"publicationDate":"2025-06-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144563461","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}

{"title":"Real-time freshness prediction for Apples and Lettuces using imaging recognition and advanced algorithms in a user-friendly mobile application","authors":"Chrysanthos Maraveas ,&nbsp;George Kalitsios ,&nbsp;Marianna I. Kotzabasaki ,&nbsp;Dimitrios V. Giannopoulos ,&nbsp;Kosmas Dimitropoulos ,&nbsp;Anna Vatsanidou","doi":"10.1016/j.atech.2025.101129","DOIUrl":"10.1016/j.atech.2025.101129","url":null,"abstract":"<div><div>Over recent decades, consumer expectations for food quality and freshness have steadily increased. To meet these standards, fresh fruits and fresh-cut vegetables in supermarkets and other commercial outlets undergo rigorous sorting processes. Quality assessments typically focus on visible characteristics such as color, ripeness, shape uniformity, defect-free skin and flesh, and texture features like firmness, toughness, and tenderness. To automate real-time quality assurance of perishable agricultural products, we have developed a user-friendly smartphone application that enables freshness assessment of apples and lettuces using RGB data at multiple stages of the supply chain. This app utilizes image recognition technology, allowing for precise freshness assessment and estimated product lifespan. Nine deep algorithms were compared in the research for image classification including Vision Transformer (ViT), Swin Transformer, Residual Networks (ResNet), EfficientNet, ConvNeXt, DeiT, MobileNetV3, MaxViT, and TNT (Transformer in Transformer). The comparison considered three metrics, including accuracy ( %), parameters (millions), and inference time (ms). Based on the findings, the MobileNetV3 was identified as the optimal deep learning architecture for the apple and lettuce classification because it maintained a good compromise between classification accuracy and mobile device resource constraints - (99.95 % and 2.5 ms for apple; 99.17 % and 2.5 million for lettuce). Such advancements offer valuable insights for policymakers, farmers, and stakeholders in making more informed decisions, thus supporting sustainable agricultural practices and improving food security across supply chains.</div></div>","PeriodicalId":74813,"journal":{"name":"Smart agricultural technology","volume":"12 ","pages":"Article 101129"},"PeriodicalIF":6.3,"publicationDate":"2025-06-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144491286","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}

{"title":"Intelligent multi-modeling reveals biological relationships and adaptive phenotypes for dairy cow adaptation to climate change","authors":"Robson Mateus Freitas Silveira ,&nbsp;Angela Maria de Vasconcelos ,&nbsp;Concepta McManus ,&nbsp;Luiz Paulo Fávero ,&nbsp;Iran José Oliveira da Silva","doi":"10.1016/j.atech.2025.101128","DOIUrl":"10.1016/j.atech.2025.101128","url":null,"abstract":"<div><div>The adaptation of animals to thermal stress involves various variables and adaptive mechanisms. These mechanisms are complex and interconnected, involving both linear and non-linear interactions and dependencies. In this study, we develop a systematic methodology with multivariate models and machine learning algorithms to (<em>i</em>) model complex patterns of relationships or multi-phenotypic differences between the thermal environment and thermoregulatory, hormonal, biochemical, hematological and productive responses; and <em>(ii</em>) identify potential associations among biological relationships that may underlie shared and specific phenotypic patterns of adaptive responses. Thirty clinically healthy multiparous lactating cows with body condition score 3–4 under the same nutritional, health and reproductive management conditions were used in the study. A simple correlation matrix revealed weak and nonexistent correlations between the variables. However, when canonical correlation analysis was used, 12 out of 15 of the canonical correlations evaluated were significant (<em>p</em> &lt; 0.05). Moderate levels of canonical correlations (0.300 ≤ <span><math><msub><mi>r</mi><mi>c</mi></msub></math></span> ≤ 0.628) and low values of squared canonical correlation (0.141 ≤ <span><math><msubsup><mi>r</mi><mi>c</mi><mn>2</mn></msubsup></math></span> ≤ 0.384) between indicators (thermal environment, thermoregulatory responses, biochemistry, hormonal profile, hematological responses and milk composition) were reported. Exceptionally, the thermal environment × biochemistry pair demonstrated notably high values (<span><math><msub><mi>r</mi><mi>c</mi></msub></math></span> = 0.8468 and <span><math><msubsup><mi>r</mi><mi>c</mi><mn>2</mn></msubsup></math></span>= 0.7171). Biological analysis formed seven distinct mechanisms, each associated with specific biological functions and climate-driven effects on physiological and productive traits. <em>1</em>) Blood traits were related to all milk components; <em>2</em>) Lipid and energy metabolism, as well as kidney function, are related to the regulation of body temperature and milk composition; <em>3</em>) Immunity and thyroid hormones are related to radiant thermal load; and <em>4</em>) Homeostasis is the organic balance maintained between thermoregulatory, hormonal, hematological, productive, and biochemical functions, which are influenced by environmental variables. Applying the random forest method to classify predictions of adaptive responses based on climatic variables showed that all thermoregulatory, hormonal, biochemical, and hematological responses are important, except for urea and T₃ concentrations, which had negative importance values. We conclude that adaptation results from an integration between energy and lipid metabolism, renal function, hormonal profile, and thermoregulatory, hematological, and productive responses. Finally, we recommend the use of multi-models to reveal the complexity o","PeriodicalId":74813,"journal":{"name":"Smart agricultural technology","volume":"12 ","pages":"Article 101128"},"PeriodicalIF":6.3,"publicationDate":"2025-06-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144519207","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}

{"title":"Construction and application of a drought classification model for tea plantations based on multi-source remote sensing","authors":"Yang Xu ,&nbsp;Yilin Mao ,&nbsp;He Li ,&nbsp;Xiaojiang Li ,&nbsp;Litao Sun ,&nbsp;Kai Fan ,&nbsp;Zhipeng Li ,&nbsp;Shuting Gong ,&nbsp;Zhaotang Ding ,&nbsp;Yu Wang","doi":"10.1016/j.atech.2025.101132","DOIUrl":"10.1016/j.atech.2025.101132","url":null,"abstract":"<div><div>In the backdrop of global climate change, drought is identified as a major natural hazard, posing a severe threat to tea production. Traditional methods for assessing drought stress in tea plants rely on manual investigation. However, this approach is time-consuming and labor-intensive. While unmanned aerial vehicle (UAV) remote sensing offers efficient alternatives, existing studies predominantly rely on single-sensor data (e.g., multispectral (MS) or thermal infrared (TIR)), overlooking the potential of multi-source fusion—especially for tea plantations. To address this gap, we propose RSDCM (Remote Sensing-based Drought Classification Model), an improved Genetic Algorithm-Backpropagation (GA-BP) combined with MS + TIR framework that optimizes initial weights and thresholds via GA's global search (hidden layer=1, neurons in the hidden layers=5, 50 generations, population size=5, NonUnifMutation operators) to escape local minima and accelerate convergence. A UAV platform equipped with MS, RGB, and TIR sensors collected multi-source data from drought-stressed tea plantations in Eastern China. The RSDCM model was benchmarked against single BP and three classical machine learning models (SVM, RF, ELM).</div><div>The study found that: (1) Multi-source data fusion outperformed single-source data, with MS + TIR achieving optimal performance (Accuracy: 0.983, Precision: 0.967-1.000, Recall: 0.967-1.000, F1-score: 0.967-1.000)—surpassing MS (Accuracy: 0.950, Precision: 0.894-1.000, Recall: 0.917-0.983, F1-score: 0.924-0.983), TIR (Accuracy: 0.925, Precision: 0.862-0.982, Recall: 0.867-0.983, F1-score: 0.889-0.967), and RGB (Accuracy: 0.904, Precision: 0.824-0.950, Recall: 0.783-0.950, F1-score: 0.847-0.950) alone. (2) The RSDCM model (accuracy: 0.983) performed better than the other four models, with high generalizability across all drought levels (F1-scores: 0.967–1.000 for severe/moderate/light/normal classes). (3) The RSDCM model could accurately classify drought stress levels in tea plantations.</div><div>Thus, RSDCM provides a novel, robust solution for UAV-based drought assessment in tea plantations, combining multi-sensor fusion and deep learning.</div></div>","PeriodicalId":74813,"journal":{"name":"Smart agricultural technology","volume":"12 ","pages":"Article 101132"},"PeriodicalIF":6.3,"publicationDate":"2025-06-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144502459","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}

{"title":"SSMR-Net and Across Feature Mapping Attention are jointly applied to the UAV imagery semantic segmentation task of weeds in early-stage wheat fields","authors":"Xinyu Mei ,&nbsp;Changchun Li ,&nbsp;Yinghua Jiao ,&nbsp;Guangsheng Zhang ,&nbsp;Longfei Zhou ,&nbsp;Xifang Wu ,&nbsp;Taiyi Cai","doi":"10.1016/j.atech.2025.101077","DOIUrl":"10.1016/j.atech.2025.101077","url":null,"abstract":"<div><div>Wheat is a critical global food crop, and its yield is significantly affected by various factors, including weeds, which can pose a major threat. Accurate identification and localization of weeds is essential for precision weeding in modern smart agriculture, with early prevention playing a key role. However, during the early growth stages, the challenge intensifies due to the significant variation in weed size, the abundance of small weeds, and the complexity of the field environment, all of which make segmentation more difficult. To address this challenge, this study combines Across Feature Mapped Attention (AFMA) with a proposed SSMR-Net model based on an improved U-Net architecture to improve weed identification. AFMA leveraged multilevel features from the original image to quantify the intrinsic relationships between large and small objects within the same category, compensating for the loss of high-level features in small target extraction and enhancing segmentation performance. SSMR-Net incorporated a multiscale feature structure by connecting the encoder and decoder with an Atrous Spatial Pyramid Pooling (ASPP) module with a small expansion rate, preserving the small target features during the information transfer and facilitating the multiscale feature extraction of weeds. The semantic differences between feature layers at the same depth were optimized through the upsampling and connection modules, whereas the encoder and decoder layers integrated an improved residual module. The skip mechanism further enabled SSMR-Net to capture features at various levels. This makes SSMR-Net maintain high segmentation performance in different complex scenarios. The combination of SSMR-Net and AFMA is more suitable for the UAV imagery semantic segmentation task of weeds in early-stage wheat fields. The experimental results demonstrated that the proposed SSMR-Net combined with AFMA achieved superior segmentation accuracy for weed and wheat identification on a custom-built wheat and weed dataset, outperforming existing models with a weed accuracy of 0.774, an IoU score of 0.696, and an mIoU of 0.865. This study presents a promising approach to precise weed identification and control in agriculture.</div></div>","PeriodicalId":74813,"journal":{"name":"Smart agricultural technology","volume":"12 ","pages":"Article 101077"},"PeriodicalIF":6.3,"publicationDate":"2025-06-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144514141","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}

{"title":"Adaptive Echo State Network for crop yield prediction incorporating Fall Armyworm dynamics","authors":"Mulima Chibuye ,&nbsp;Jackson Phiri ,&nbsp;Phillip Nkunika","doi":"10.1016/j.atech.2025.101123","DOIUrl":"10.1016/j.atech.2025.101123","url":null,"abstract":"<div><div>Agricultural productivity worldwide is threatened by invasive pests, notably the Fall Armyworm (FAW, <em>Spodoptera frugiperda</em>), which has devastated maize yields across Africa and Asia since 2016. To support precision pest management, we developed an adaptive Echo State Network (ESN) that predicts annual maize yield while accounting for FAW pressure. We compiled a 15-year (2010–2024) monthly dataset combining satellite vegetation indices, in-field weather, soil chemistry readings, and FAW surveillance counts. FAW severity is quantified on a 0–100 scale by blending trap counts (40 %) and larval density (60 %) per month. First, the ESN is trained on all available data to predict crop yields based on environmental features. We then apply isotonic regression to map pest infestation levels to the ESN's residual over-predictions, producing a monotonic penalty curve. This curve quantifies yield losses at different pest pressures. During prediction, we apply this learned penalty to the raw ESN output, adjusting yield estimates to account for pest damage without altering the original ESN model. In cross-validation, the FAW-aware ESN achieves an R² of ∼0.55 and reduces prediction errors by up to 67 % versus unpenalized baselines, closely capturing observed yield reductions exceeding 20 % during severe outbreaks. The model outperforms standard regression and deep neural network approaches by similar margins. It guides farmers in targeting interventions to high-risk zones, reducing pesticide use and operational costs. These results highlight its value as an early-warning tool for targeted interventions that minimize chemical inputs and optimize resource allocation. Ongoing field validations will evaluate its scalability and practical impact in FAW-affected maize production regions.</div></div>","PeriodicalId":74813,"journal":{"name":"Smart agricultural technology","volume":"12 ","pages":"Article 101123"},"PeriodicalIF":6.3,"publicationDate":"2025-06-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144481625","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}

{"title":"Prediction of winter wheat nitrogen nutrition index using high-resolution satellite and machine learning","authors":"Po-Ting Pan ,&nbsp;Yamine Bouzembrak ,&nbsp;Miguel Quemada ,&nbsp;Bedir Tekinerdogan","doi":"10.1016/j.atech.2025.101119","DOIUrl":"10.1016/j.atech.2025.101119","url":null,"abstract":"<div><div>Wheat (<em>Triticum aestivum</em> L.) is the most important cereal crop grown in Spain, and Spain is one of the top wheat-producing countries in EU. Precision fertilization, which customizes the fertilizer dosage based on the variability of the field, is important for the environment, food security, and farmers’ finances. To provide the fertilizer prescription, assessing crop nitrogen (N) status is required to make site-specific fertilizer applications. Among the nitrogen status index, the nitrogen nutrition index (NNI) is considered the most reliable to monitor the N status. Prior studies have used satellite, UAV, or spectral sensors with mostly adopting linear regression methods to predict NNI. However, no study has investigated the potential of using high-resolution satellite and environmental data with machine learning (ML) to predict winter wheat NNI. Therefore, this study integrated PlanetScope satellite images with weather data while adopting three ML algorithms, including random forest (RF), support vector machine (SVM), and artificial neural network (ANN) to predict NNI in Spain from 2018 to 2019. The results showed that RF outperformed the SVM and ANN models with an accuracy of 77.08 % and a precision of 0.78. This study also demonstrated weather data improved model performance across all three algorithms with the highest accuracy of 79.12 % in the RF algorithm. Among all three algorithms, the elongation period outperformed the flowering period and across the entire period with an accuracy of 81.25 - 87.5 % and a precision of 0.5 - 0.78. In the end, the N status diagnostic map was generated to reflect the nitrogen requirement and provide a decision support tool for farmers before the fertilizer application. The proposed methodology in this paper can be extended to different crops and different regions for NNI prediction.</div></div>","PeriodicalId":74813,"journal":{"name":"Smart agricultural technology","volume":"12 ","pages":"Article 101119"},"PeriodicalIF":6.3,"publicationDate":"2025-06-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144491285","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}