Smart agricultural technology最新文献

{"title":"Transformer-based and band-selected models for UAV hyperspectral wheat disease classification","authors":"Alireza Sanaeifar ,&nbsp;Shahryar Kianian ,&nbsp;Ruth Dill-Macky ,&nbsp;Susan Reynolds ,&nbsp;Matthew J Moscou ,&nbsp;Rebecca D. Curland ,&nbsp;James Anderson ,&nbsp;Matthew N. Rouse ,&nbsp;Ce Yang","doi":"10.1016/j.atech.2025.101714","DOIUrl":"10.1016/j.atech.2025.101714","url":null,"abstract":"<div><div>Rapid, accurate diagnosis of wheat diseases is essential for targeted crop management and food-security resilience, yet field-scale monitoring remains constrained by labor-intensive visual surveys. Here, we couple unmanned aerial vehicle (UAV) hyperspectral imaging (395–903 nm, 240 bands) with advanced machine learning to form an end-to-end classification pipeline. We first assembled a curated single-pathogen UAV spectral library—3,105 plot-level spectra representing five wheat diseases; bacterial leaf streak, Fusarium head blight and the three major rusts—collected from controlled field trials at US midwestern research stations. Principal-component analysis revealed disease-specific reflectance patterns in both the visible and near-infrared domains, guiding downstream model design. We introduce BandWiseTransformer, a one-dimensional self-attention network tailored to hyperspectral vectors. Automated hyperparameter optimization with Optuna yielded an architecture (128-dimensional embedding, four heads, four encoder layers) that achieved 98 % test accuracy and class-average precision, recall and F1-score of 0.98, decisively outperforming conventional baselines. By analyzing the network’s embedding and classifier weights, we derived a wavelength-importance metric and identified the 10 most informative bands. Random Forest and XGBoost models trained solely on this compressed feature set retained respectable accuracies of 0.86 and 0.88, while slashing dimensionality by 96 % and inference time by two orders of magnitude. SHapley Additive exPlanations (SHAP) analysis confirmed that the selected green, red-edge and near-infrared bands—illustratively 492, 715, 780, 839 and 901 nm—capture biochemical and structural cues linked to pigment loss, red-edge shifts and canopy water status. Thus, transformer attention not only maximises full-spectrum accuracy but also exposes a minimal band subset that underpins lightweight, interpretable classifiers. This study delivers (i) a curated UAV hyperspectral dataset, (ii) a high-accuracy spectral transformer architecture, and (iii) a transparent path from deep-model insights to cost-effective multispectral sensor design. Together, these advances enable real-time scouting of wheat diseases and establish a transferable workflow for hyperspectral crop-health monitoring in diverse pathosystems.</div></div>","PeriodicalId":74813,"journal":{"name":"Smart agricultural technology","volume":"13 ","pages":"Article 101714"},"PeriodicalIF":5.7,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145925498","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":"A study on blueberry variety classification based on the HMS-ResNeXt50 model","authors":"Rongli Gai ,&nbsp;Jiyun Zhang ,&nbsp;Mingyue Gao ,&nbsp;Lingyan Hu ,&nbsp;Guohui Xu ,&nbsp;Xiaoyu Nie","doi":"10.1016/j.atech.2025.101734","DOIUrl":"10.1016/j.atech.2025.101734","url":null,"abstract":"<div><div>Currently, blueberry variety classification faces two major challenges: first, the traditional manual identification method is inefficient, and the results are greatly affected by subjective experience, making it difficult to ensure consistency; Second, the classification approach based on spectral characteristics depends too much on manually crafted features and linear models, failing to handle interference from complex environmental variables and meet the requirements of nonlinear classification scenarios. To address these issues, we propose a blueberry leaf classification method based on the HMS-ResNeXt50 network. In this study, &gt;2700 blueberry leaf images were collected to create a dataset, and image features were enhanced through image preprocessing and data augmentation. The HMS-ResNeXt50 incorporates the designed multi-scale enhanced aggregation convolution and multi-scale depthwise convolution parallel feature extraction block. It also integrates SE attention to strengthen the feature extraction capability. We evaluated our method on the constructed dataset. Experiments show that HMS-ResNeXt50 performs well in blueberry leaf classification, achieving a final accuracy of 0.9276±0.0045, precision of 0.9345±0.0032, recall of 0.9291±0.0022, and F1-score of 0.9318±0.0019.This study realizes the automatic recognition of blueberry leaves based on deep learning technology, offering not only valuable references for blueberry variety identification but also providing a novel approach to address this task.</div></div>","PeriodicalId":74813,"journal":{"name":"Smart agricultural technology","volume":"13 ","pages":"Article 101734"},"PeriodicalIF":5.7,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145925487","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":"Optimizing artificial lighting for convolutional neural network-based crop monitoring with low-cost RGB imaging in indoor cultivation","authors":"Matteo Landolfo,&nbsp;Fabio Perotti,&nbsp;Alessandro Pistillo,&nbsp;Giuseppina Pennisi,&nbsp;Giorgio Gianquinto,&nbsp;Francesco Orsini","doi":"10.1016/j.atech.2025.101677","DOIUrl":"10.1016/j.atech.2025.101677","url":null,"abstract":"<div><div>This study investigated the effect of different red:blue (R:B) spectral light ratios on the performance of a multi-task convolutional neural network (CNN) model developed for the automatic classification of four horticultural species and their corresponding phenological stages under controlled artificial lighting conditions. The model was trained and tested using RGB images acquired under five distinct spectral treatments (R:B 1, 3, 5, 7, and 9), and its performance was evaluated using accuracy, precision, recall, F1-score, and Matthews correlation coefficient (MCC). For species classification, the best results were obtained with an R:B 1, achieving an accuracy of 86 %, precision of 87 %, recall of 85 %, F1-score of 85 %, and MCC of 0.81. In terms of phenological stage classification, the highest performance was observed at R:B 3 and R:B 5, both yielding 93 % accuracy and F1-score, precision and recall above 92 %, and an MCC of 0.86. These findings demonstrate that the multi-task CNN model is capable of learning robust and generalizable representations, maintaining high classification performance even under non-optimal spectral conditions. The integration of optimized artificial lighting with intelligent classifiers proves to be a strategic approach for automated monitoring systems in indoor and precision agriculture. Future research should explore the impact of additional spectral components (e.g., green or far-red wavelengths) and the adoption of more advanced neural architectures to further enhance the system’s robustness and scalability.</div></div>","PeriodicalId":74813,"journal":{"name":"Smart agricultural technology","volume":"13 ","pages":"Article 101677"},"PeriodicalIF":5.7,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145925488","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":"CSM-YOLO: An integrated lightweight framework for UAV-based detection and geospatial analysis of missing cotton seedlings","authors":"Xu Wang ,&nbsp;Nueraili Aierken ,&nbsp;Xuelin Zhou ,&nbsp;Zhixin Yao ,&nbsp;Bo Yang ,&nbsp;Yongke Li ,&nbsp;Xiangping Feng","doi":"10.1016/j.atech.2025.101751","DOIUrl":"10.1016/j.atech.2025.101751","url":null,"abstract":"<div><div>Accurate monitoring of cotton seedling establishment is essential for optimizing yield and field management in precision agriculture. Gaps caused by germination failure or uneven mechanical sowing can substantially reduce final yield and complicate subsequent field operations. This study aimed to develop a real-time, accurate, and scalable framework for monitoring missing cotton seedling establishment and quantifying missing-seedling gaps from UAV imagery. We propose CSM-YOLO, an end-to-end integrated framework comprising three core components: (i) a lightweight yet high-performance detection model (CSM-YOLO), in which the CA-StarNet backbone enables efficient feature encoding, the multi-scale C3k2-RAB module enhances fine-grained seedling representation, and the adaptive APT-TAL label assignment strategy improves feature–anchor matching accuracy, with the three components jointly achieving an optimal balance among accuracy, robustness, and inference efficiency; (ii) two complementary pixel-to-geographic coordinate mapping strategies, including a novel linear interpolation method and an enhanced collinearity equation method, each offering distinct advantages in terms of computational efficiency and georeferencing accuracy; and (iii) a regional grading and visualization module for missing-seedling rates that supports both quantitative assessment and spatial interpretation within a unified workflow. CSM-YOLO achieved a mean average precision ([email protected]) of 97.23%, with class-specific precisions of 97.87% for cotton seedlings and 96.89% for missing-seedling holes, significantly outperforming mainstream models (Faster R-CNN, SSD, YOLOv3-tiny, YOLOv5n, YOLOv8n, YOLOv10n, YOLOv11n, and YOLOv11s). The two coordinate-mapping strategies yielded mean localization errors of 0.155 m and 0.004 m, respectively, when validated against orthophoto reference points. The framework provides high-precision detection, efficient georeferencing, and intuitive visualizations of seedling missing patterns. It explores the feasibility of an integrated pipeline that links detection, localization, and visualization for UAV-based cotton field monitoring. By coupling these components into a coherent workflow, the proposed system enables high-throughput quality assessment, precise reseeding planning, and accurate identification of weed-infested zones, thereby offering a robust and scalable solution for data-driven decision-making in precision cotton production.</div></div>","PeriodicalId":74813,"journal":{"name":"Smart agricultural technology","volume":"13 ","pages":"Article 101751"},"PeriodicalIF":5.7,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145925489","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":"Detection method of broken grains and impurities in harvested soybeans using feature wavelength selection and MobileNetV4-Unet-SGCPNet hybrid network","authors":"Chengqian Jin ,&nbsp;Gong Cheng ,&nbsp;Zhichang Chang ,&nbsp;Man Chen ,&nbsp;TengXiang Yang ,&nbsp;Yinyan Shi ,&nbsp;Xiaobin Gai","doi":"10.1016/j.atech.2025.101749","DOIUrl":"10.1016/j.atech.2025.101749","url":null,"abstract":"<div><div>To address the challenges of hyperspectral data redundancy, small-target segmentation difficulty, and insufficient model real-time performance in high-precision online detection of foreign matter and kernel breakage in machine-harvested soybeans, this study proposes a collaborative detection method based on \"feature wavelength optimization + MobileNetV4-Unet-SGCPNet hybrid network\". First, 18 key feature bands were screened from 400 - 1000 nm hyperspectral data using successive projection algorithm (SPA) and competitive adaptive reweighted sampling (CARS), constructing a multi-source feature spectral image dataset. Subsequently, a MobileNetV4-Unet-SGCPNet hybrid network was designed, with a lightweight MobileNetV4 as the encoder, combined with the symmetric encoder-decoder structure of Unet and the spatial detail-guided context propagation module (SGCP) to achieve high-precision segmentation of broken grains, complete grains, and impurities. Finally, pixel-wise voting was employed to fuse multi-band feature information, enhancing the model’s generalization capability. The results demonstrate that: on the test set, the model achieves an average intersection - over - union of 89.69 % for soybean component recognition, with a mean precision average of 94.55 %, a mean precision of 93.63 %, a frame rate of 4.77 FPS, a parameter count of only 2.86 MB, and a computational load of 35.78 GFLOPs. Compared to mainstream models, this method reduces parameters by 97.2 % and computational cost by 97.8 %, while the average intersection - over - union drops by only 6.2 %, with a frame rate improvement of over 5 times, striking a significant balance between detection accuracy and real-time performance. Cross-variety and cross-device validations further confirm that the model effectively adapts to morphological and spectral variations across different soybean varieties, exhibiting strong generalization ability. This study provides a core algorithmic foundation for online monitoring systems of intelligent harvester operation quality, offering critical support for enhancing the commercial value of machine-harvested soybeans and advancing the intelligence level of agricultural machinery.</div></div>","PeriodicalId":74813,"journal":{"name":"Smart agricultural technology","volume":"13 ","pages":"Article 101749"},"PeriodicalIF":5.7,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145925491","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":"Data mining for evaluating animal performance from weighing platform big data","authors":"Alex S.C. Maia ,&nbsp;Gustavo A.B. Moura ,&nbsp;Vinicius F.C. Fonsêca ,&nbsp;Bruno R. Simão ,&nbsp;Jessica O. Gusmão ,&nbsp;Hugo F.M. Milan ,&nbsp;Kifle G. Gebremedhin ,&nbsp;Robert J. Collier ,&nbsp;Rodrigo D.L. Pacheco ,&nbsp;Izabelle A.M.A. Teixeira","doi":"10.1016/j.atech.2025.101758","DOIUrl":"10.1016/j.atech.2025.101758","url":null,"abstract":"<div><div>Precision livestock farming technologies, such as IoT-enabled weighing platforms, generate big data that require rigorous processing to ensure correctness and integrity. However, literature lacks a reliable procedure for automatically filtering misreadings from weighing platform big data, hindering accurate evaluation of performance indicators. This study addresses this gap through developing and validating a data mining procedure consisting in four steps: removes outliers based on 1) shrunk body weight (<em>SBW</em>), 2) discrepancies between weighing platform measurements, 3) estimation of initial and final live body weight (<em>BW</em>), and 4) deviations between measured and estimated live <em>BW</em> from a Generalized Additive Model (GAM). Data were collected from 12 experiments (1152 steers, 583,321 samples) conducted in Campanelli Innovation Center. The data mining procedure effectively removed outliers from loss of calibration, partial or even multiple animals standing on the platform. When compared with previous methodologies, our procedure was 20x more accurate. Mined data were used to fit GAM to predict <em>BW</em>. The average daily gain calculated from mined data showed an error (–0.02 ± 0.74%) 40x more accurate than using raw data (7.99 ± 29.42%) or previous methodologies and showed no statistical difference with measured data. Considering that our proposed methodology needs <em>SBW</em> and its measurement is stressful and lead to performance loss, an extreme gradient boosting regression model was trained to predict initial and final SBW and were highly accurate (RMSE: 0.0216 kg; R²: 0.9954). The proposed data mining procedure improved data correctness and integrity, generating high-quality datasets that can be used for early detection of animal disease, evaluating impacts on performance from new feeding additives, and supporting data-driven management decisions. In addition, the proposed procedure can be easily implemented in similar set ups.</div></div>","PeriodicalId":74813,"journal":{"name":"Smart agricultural technology","volume":"13 ","pages":"Article 101758"},"PeriodicalIF":5.7,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145925637","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":"UAV remote sensing for rice false smut detection via improved YOLOv12n","authors":"Haiyong Weng ,&nbsp;Leilei Su ,&nbsp;Yuanpeng Liu ,&nbsp;Geqiang Lu ,&nbsp;Xintong Cao ,&nbsp;Jie Lin ,&nbsp;Wenjie Xu ,&nbsp;Hairong Luo ,&nbsp;Zuxin Cheng ,&nbsp;Dapeng Ye","doi":"10.1016/j.atech.2025.101756","DOIUrl":"10.1016/j.atech.2025.101756","url":null,"abstract":"<div><div>Rice false smut (RFS) has become widespread in major rice-growing regions. It not only reduce rice production, but more importantly, it produces toxins on panicles—posing a significant threat to food safety. Rapidly and accurately detecting RFS plays a key role in ensuring rice production. This study proposes a lightweight detection model named HSFF-YOLO, which is based on an improved YOLOv12n and utilizes unmanned aerial vehicle (UAV) remote sensing images. To address the challenge of small target detection during model development, the original loss function and Neck structure were replaced with the Shape-NWD loss function and the High-level Selective and Feature Fusion (HSFF) multi-scale feature fusion structure, respectively. This replacement effectively enhances the model’s detection accuracy through semantic screening and multi-scale feature fusion. To further highlight the lesion information of RFS, the Convolutional Block Attention Module-Parallel (CBAM-P) was incorporated to emphasize semantic and spatial information. Additionally, the lightweight DySample (Upsampling by Dynamic Sampling) upsampling module was introduced to reduce the model’s computational resource overhead. The results showed that the HSFF-YOLO model achieved an average accuracy of 80.7 %, with parameters of 1.89 M, a Floating Point Operations (FLOPs) of 5.6 G, and a model size of 4.09 MB. Compared with YOLOv12n, it has decreased 25 % and 21 % of parameters and model size, respectively, while still presented relatively better mAP0.5 %, precision and recall, with approximate 3 %, 3.4 % and 2.4 % improvement. This study also conducted evaluations of the severity levels of RFS in the two regions. The results showed that RFS in the Longyan region was significantly more severe. This study indicated that UAV imaging coupled with HSFF-YOLO has excellent potential for RFS detection and completed the entire process from the rapid detection of RFS to disease severity assessment.</div></div>","PeriodicalId":74813,"journal":{"name":"Smart agricultural technology","volume":"13 ","pages":"Article 101756"},"PeriodicalIF":5.7,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145925564","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":"Drivers, barriers and grower perspectives of innovation adoption in the UK controlled environment agriculture sector","authors":"Iona Y Huang ,&nbsp;Andrew M Beacham ,&nbsp;Laura H Vickers ,&nbsp;Hairong Mu ,&nbsp;Ourania Tremma ,&nbsp;Elias Maritan ,&nbsp;Magdalena Kaczorowska-Dolowy ,&nbsp;James M Monaghan","doi":"10.1016/j.atech.2025.101748","DOIUrl":"10.1016/j.atech.2025.101748","url":null,"abstract":"<div><div>Controlled environment agriculture (CEA), which comprises both greenhouse and vertical farming, is a multi-billion-pound global industry contributing significantly to fruit and vegetable production. CEA allows growers to produce crops reliably in otherwise unfavourable climatic conditions and permits extended growing seasons. CEA is ideally placed to exploit the technological and digital revolution in agriculture for precision control of crop growth conditions, health, quality and yield. Despite the value of the sector, there is a paucity of data regarding drivers and barriers, adoption rates and grower opinion of implementing advanced technologies in commercial CEA production. We combined a Quick Scoping Review of global CEA technology literature comprising 3679 primary studies with a Delphi study of UK CEA growers and technology providers (25 and 18 participants in rounds one and two, respectively) to determine CEA technology research focus, stakeholder priorities and concerns. Comparison of categorised technology types between these approaches was used to assess the degree of synergy between the two. This revealed a global CEA technology research focus on modelling/simulation, energy, lighting and sensors. Meanwhile, the most used technologies in UK CEA were environmental control, lighting and nutrient application. This illustrates a fair degree of connect between research focus and industry requirements. However, some grower priorities, including nutrition and robotics, were under-represented in the literature. UK grower behaviour remains optimistic regarding the opportunities technology such as alternative energy sources and automation can offer CEA, including reduced costs, improved efficiency and crop quality, extended supply season and mitigation of import risk. This must, however, be balanced with threats including operational and capital costs, which are viewed as the most significant barriers to technology innovation uptake.</div></div>","PeriodicalId":74813,"journal":{"name":"Smart agricultural technology","volume":"13 ","pages":"Article 101748"},"PeriodicalIF":5.7,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145925565","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":"Classification of soybean genotypes based on physiological clustering (PCA + k-means) integrated with VIS-NIR hyperspectral data and machine learning models","authors":"Regimar Garcia dos Santos ,&nbsp;Dthenifer Cordeiro Santana ,&nbsp;Larissa Pereira Ribeiro Teodoro ,&nbsp;Cid Naudi Silva Campos ,&nbsp;Fabio Henrique Rojo Baio ,&nbsp;Carlos Antônio da Silva Junior ,&nbsp;Elber Vinicius Martins Silva ,&nbsp;Luan Pereira de Oliveira ,&nbsp;Paulo Eduardo Teodoro","doi":"10.1016/j.atech.2025.101702","DOIUrl":"10.1016/j.atech.2025.101702","url":null,"abstract":"<div><div>Given the wide variety of genotypes, classifying them based on physiological traits can aid breeding efforts. This study aimed to identify the most effective machine learning algorithms and input variables for classifying soybean genotypes according to physiological features. A field experiment was conducted using a random complete block design with three replications, testing 32 soybean genotypes. Multispectral and hyperspectral reflectance data were collected and grouped into 20 representative wavelength ranges. The physiological traits measured included net photosynthesis, internal CO₂ concentration, stomatal conductance, and transpiration rates. Genotype groups based on physiological performance were classified with the k-means algorithm and principal component analysis, resulting in two clusters. K-means is an algorithm that groups data into k similar groups. It does this by finding patterns and placing each sample in the group whose center (centroid) it is closest to. These clusters served as output variables in machine learning models, with input variables including wavelengths and spectral band averages. A hyperspectral sensor, was employed to record leaf reflectance at wavelengths ranging from 450 to 824 nm in the laboratory under artificial lighting. The algorithms tested were artificial neural networks, J48 decision trees, REPTree, support vector machines (SVM), random forest, and logistic regression (LR) as a control. Model accuracy was assessed using correct classification percentage and F-score. SVM and RL stood out with accuracies above 0.6 in classifications for CC and an F-score exceeding 0.75. When using spectral bands as predictors, both showed similar performance, but with wavelengths as predictors, WL becomes a robust input for the models because its complete spectrum information provides important data for group classifications.</div></div>","PeriodicalId":74813,"journal":{"name":"Smart agricultural technology","volume":"13 ","pages":"Article 101702"},"PeriodicalIF":5.7,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145925720","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":"Dynamic monitoring of maize field vegetation cover using sentinel-1 and sentinel-2 data and transfer learning algorithms","authors":"Hong Sun,&nbsp;Weichao Liu,&nbsp;Miao Wang,&nbsp;Jinjin Li,&nbsp;Ruiyan Wang","doi":"10.1016/j.atech.2025.101711","DOIUrl":"10.1016/j.atech.2025.101711","url":null,"abstract":"<div><div>Maize vegetation cover (FVC) is a key indicator for assessing crop growth and resource utilization, and is of great importance for yield estimation, precision agriculture management, and sustainable agricultural development. To address the issue of missing optical images during critical growth stages of maize due to cloudy and rainy weather, this study constructed an FVC prediction model for Sentinel-2 imagery based on multi-temporal Sentinel-1 and Sentinel-2 data, integrating a pixel dichotomy model, temporal variation features, and a transfer learning algorithm. The model enables the monitoring of FVC dynamics during cloudy–rainy periods, and the results were validated using UAV imagery. The findings indicate that introducing temporal variation features as prior knowledge effectively alleviates the dependence of transfer learning on large-scale samples, supporting FVC prediction and mapping for summer maize under small-sample conditions. The predicted images show strong consistency with real Sentinel-2 imagery in spatial patterns and vegetation growth dynamics, achieving an overall correlation of 0.75. Among different growth stages, the highest prediction accuracy occurred from the jointing to tasseling stage (R² = 0.55), while lower accuracies were observed for the jointing–booting and booting–tasseling stages (R² = 0.35 and 0.44, respectively). The study also revealed that periods with larger FVC fluctuations corresponded to higher prediction accuracy. The proposed method performed well for both spring maize and summer maize. The study verifies the feasibility of integrating microwave and optical remote sensing data with transfer learning algorithms for estimating maize FVC. It reveals how growth stages influence prediction performance and provides a practical solution for dynamic crop cover monitoring under cloudy and rainy weather conditions, enabling stable crop monitoring in regions with frequent optical data gaps.</div></div>","PeriodicalId":74813,"journal":{"name":"Smart agricultural technology","volume":"13 ","pages":"Article 101711"},"PeriodicalIF":5.7,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145925724","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}