Computers and Electronics in Agriculture最新文献

{"title":"Improving multi-step dissolved oxygen prediction in aquaculture using adaptive temporal convolution and optimized transformer","authors":"Kaixuan Shao ,&nbsp;Daoliang Li ,&nbsp;Hao Tang ,&nbsp;Yonghui Zhang ,&nbsp;Bo Xu ,&nbsp;Uzair Aslam Bhatti","doi":"10.1016/j.compag.2025.110329","DOIUrl":"10.1016/j.compag.2025.110329","url":null,"abstract":"<div><div>Accurate dissolved oxygen (DO) prediction is crucial for optimizing aquaculture efficiency. However, existing forecasting methods often struggle to capture periodic patterns, model complex feature dependencies, and maintain high accuracy in multi-step predictions. To address these challenges, this study proposes an enhanced Transformer-based model designed to improve both prediction accuracy and stability. The model first integrates an Adaptive Temporal Convolutional Network (ATCN) to extract periodic patterns and local temporal features from time-series data. Then, a Transformer encoder with linear attention mechanisms and relative position encoding is employed to enhance feature extraction and sequence modeling. To further strengthen temporal dependency learning, the conventional decoder is replaced with a modified Gated Recurrent Unit (GRU), and a linear regression-based error correction mechanism is introduced to refine multi-step forecasting accuracy. Experimental results on Public Dataset 3 demonstrate that the proposed model achieves an average MAE of 1.624 and RMSE of 2.254, reflecting improvements of 45.06% and 40.57%, respectively, compared to the baseline Transformer model. These findings highlight the model’s ability to effectively capture complex temporal dependencies, significantly enhancing DO prediction accuracy and robustness in multi-step forecasting tasks. Additionally, the model demonstrates strong performance in pH prediction, underscoring its potential for multi-parameter water quality forecasting. This work provides valuable insights for improving water quality management and mitigating risks in aquaculture.</div></div>","PeriodicalId":50627,"journal":{"name":"Computers and Electronics in Agriculture","volume":"235 ","pages":"Article 110329"},"PeriodicalIF":7.7,"publicationDate":"2025-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143740122","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Development of an autonomous navigation system for orchard spraying robots integrating a thermal camera and LiDAR using a deep learning algorithm under low- and no-light conditions","authors":"Ailian Jiang ,&nbsp;Tofael Ahamed","doi":"10.1016/j.compag.2025.110359","DOIUrl":"10.1016/j.compag.2025.110359","url":null,"abstract":"<div><div>Pesticide spraying is an important part of agricultural production that directly affects crop yields. However, with the agricultural labor force aging globally, conventional spraying methods face challenges due to the shortage of skilled operators in orchard management. Moreover, conventional methods usually waste a large amount of pesticides, which not only increase the production cost but also results in severe environmental pollution from pesticide residues. Furthermore, it affects food safety and the ecological balance. Therefore, this study proposes a new pesticide spraying robot that can attract pests to approach a specific spraying device via light and pheromones, thus improving the accuracy of pesticide application and reducing the amount of unnecessary spraying. For the spraying robot to work properly at night when insects are active, the spraying system needs to have the ability to navigate autonomously without being affected by light. Therefore, this study uses a thermal camera and light detection and ranging (LiDAR) as sensors for navigation, target detection and image segmentation via YOLACT (You Only Look At CoefficienTs) deep learning and fuses accurate distance data from LiDAR to realize real-time navigation of the vehicle according to the position of the trees in the orchard. This method can ensure accurate navigation of the vehicle in a dense canopy orchard environment and can enable the vehicle to operate safely under low-light and no-light conditions. The real-time navigation system proposed in this study was tested during the day and night, first in an artificial tree orchard and then in a real orchard. The experimental results revealed that in the artificial tree orchard, the image segmentation mean average precision (mAP) of the box was 83.74 %, that of the mask was 81.4 %, and the average positional error from the target travel path was 0.21 m. In the real orchard, the image segmentation mAP of the box was 62.03 %, that of the mask was 58.82 %, and the average positional error was 0.20 m. The system exhibited good stability under different light conditions, including low light and no light, in orchards and provides a solution for the development of night-time applications to control insects with reduced pesticide contents.</div></div>","PeriodicalId":50627,"journal":{"name":"Computers and Electronics in Agriculture","volume":"235 ","pages":"Article 110359"},"PeriodicalIF":7.7,"publicationDate":"2025-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143747600","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Identification of wheat kernel vitreousness by hyperspectral imaging: Comparing the Visible, Vis-NIR and SWIR range","authors":"Gözde Özdoğan,&nbsp;Aoife Gowen","doi":"10.1016/j.compag.2025.110361","DOIUrl":"10.1016/j.compag.2025.110361","url":null,"abstract":"<div><div>Vitreousness serves as a crucial visual indicator of grain hardness and is of paramount importance in the wheat industry due to its substantial influence on both milling and baking quality. Consequently, it is regarded as a fundamental criterion for assessing wheat quality and determining its market value. This study evaluates the efficacy of hyperspectral imaging (HSI) in classifying the grains of thirty-six wheat varieties as either vitreous or non-vitreous, focusing on classification performance across different spectral regions, including Visible (Vis), Visible-Near Infrared (Vis-NIR), and Short-Wave Infrared (SWIR). To achieve this, Support Vector Machines (SVM), Linear Discriminant Analysis (LDA), and Artificial Neural Networks (ANN) were utilised to classify grains according to their vitreousness. The results revealed that vitreous kernels were more readily classified than non-vitreous kernels, with classification accuracies of 93.01 % and 83.13 %, respectively. The highest F1 score for the test set, 85.26 %, was attained in the Vis-NIR range by SVM. Region of interest (ROI) selection improved non-vitreous classification by up to 3 %, particularly in the Vis and Vis-NIR regions. Furthermore, five critical wavelengths (540, 636, 476, 588, and 489 nm) in the Vis range were identified using the Minimum Redundancy Maximum Relevance (mRMR) approach. Notably, the reduced set of wavelengths yielded classification accuracies comparable to those obtained using the full spectrum, achieving an accuracy of 93.56 % for vitreous grains and 77.90 % for non-vitreous grains. These findings highlight the potential of HSI, particularly within the Vis region, for the non-destructive classification of wheat grain vitreousness, with colour information emerging as a vital factor in the classification process.</div></div>","PeriodicalId":50627,"journal":{"name":"Computers and Electronics in Agriculture","volume":"235 ","pages":"Article 110361"},"PeriodicalIF":7.7,"publicationDate":"2025-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143740124","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"YOLO-CMST: Towards accurate pineapple flowering induction using YOLO-based models with the Cross Multi-Style Translator","authors":"Kuang-Yueh Pan ,&nbsp;Wan-Ju Lin ,&nbsp;Jian-Wen Chen ,&nbsp;Yi-Hong Lin","doi":"10.1016/j.compag.2025.110315","DOIUrl":"10.1016/j.compag.2025.110315","url":null,"abstract":"<div><div>Recently, You Only Look Once (YOLO) models have been widely used in agricultural applications, such as precise flowering induction for pineapples. These models help ensure uniform maturity, optimize harvest schedules, and improve overall quality. However, YOLO models face challenges, particularly the extra effort required to collect diverse datasets and manually label them if we want to enhance model performance. To address these challenges, we propose YOLO-CMST, a solution that integrates the Cross Multi-Style Translator (CMST) and the Intermediate Domain Transformation (IDT) algorithm. This module generates synthetic pineapple images in a variety of styles while ensuring the position of the pineapple core aligns with the original images. As a result, it allows the original labeled files to be reused for synthetic pineapple images, eliminating the need for additional data collection or manual labeling during training. Based on this, we developed a self-propelled pineapple flowering spray machine that autonomously detects pineapple cores using YOLO-CMST and directs the nozzles for precise spraying. Furthermore, to achieve the practical application of the proposed framework, we deployed the model on low-cost, real-time computing systems, such as the Intel Neural Compute Stick (NCS). Experimental results demonstrate that the proposed detection system provides adequate speed for real-world field applications, ensuring both efficiency and reliability in spraying floral inducers. Additionally, the performance of YOLO models can be effectively improved by over 7% in F1-score and 4% in mean Intersection over Union (mIoU) with CMST, while maintaining the same size of the training dataset.</div></div>","PeriodicalId":50627,"journal":{"name":"Computers and Electronics in Agriculture","volume":"235 ","pages":"Article 110315"},"PeriodicalIF":7.7,"publicationDate":"2025-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143739863","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"UAVs-UGV cooperative boom sprayer system based on swarm control","authors":"Yuli Chen ,&nbsp;Zibo Liu ,&nbsp;Zifeng Xu ,&nbsp;Jianqin Lin ,&nbsp;Xianlu Guan ,&nbsp;Zhiyan Zhou ,&nbsp;Dateng Zheng ,&nbsp;Andrew Hewitt","doi":"10.1016/j.compag.2025.110339","DOIUrl":"10.1016/j.compag.2025.110339","url":null,"abstract":"<div><div>To increase the boom width of boom sprayers without adding to the overall weight, while ensuring the stability of long-span booms and improving the efficiency of boom sprayers, this paper proposes a UAVs-UGV(Unmanned Aerial Vehicle, Unmanned Ground Vehicle) Cooperative Boom Sprayer System(UCBSS) based on swarm control. The UCBSS integrates the high payload capacity of UGVs with the high maneuverability and terrain-independent characteristics of UAVs, simplifying the boom structure. Multi-rotor UAVs form a UAV swarm, which segments and suspends the boom, collaborating with the UGV to complete the spraying operation. A prototype with three UAVs and one UGV was developed, featuring a unilateral boom width of 21 m. To meet the operational requirements of the UCBSS, PD(Proportional-Derivative) feedback control is employed to achieve segmented boom balance, utilizing IMUs(Inertial Measurement Unit) installed on each boom section and RTK(Real Time Kinematic) positioning modules mounted on the UGV and UAVs. For UAVs-UGV cooperative motion control, the Adaptive Feedforward Compensation PD Feedback(AFCPF) control method is designed to control the UAV swarm. Reinforcement learning is used to train and optimize the control parameters. Field test were conducted to validate the UCBSS. The results show that in terms of boom balance, the average roll angle of the entire boom is 0.014 rad, with an average standard deviation of 0.007 rad, demonstrating high stability and mitigating the impact of boom elongation. Regarding UAVs-UGV cooperative motion, when using the proposed AFCPF control method, the maximum tracking error of the three UAVs is 0.204 m, representing a 68.3 % reduction compared to the PD control method. The overall average tracking error of the three UAVs is 0.109 m, a reduction of 60.6 % compared to the PD control method. The standard deviations are 0.030 m, 0.038 m, and 0.032 m, respectively, representing reductions of 55.2 %, 66.4 %, and 55.6 %, with an overall reduction of 59.1 %, verifying the effectiveness and stability of the proposed control method. The UCBSS proposed in this paper features a wide spraying span, simple structure, high operational stability, and easy scalability. Against the backdrop of rapid advancements in UAV and related technologies, it provides a novel approach to the development of high-efficiency, wide-span boom sprayers.</div></div>","PeriodicalId":50627,"journal":{"name":"Computers and Electronics in Agriculture","volume":"235 ","pages":"Article 110339"},"PeriodicalIF":7.7,"publicationDate":"2025-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143746276","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Real-time detection system of garlic clove bud for garlic clove orientation metering device based on machine vision","authors":"Yongzheng Zhang ,&nbsp;Jingling Song ,&nbsp;Long Zhou","doi":"10.1016/j.compag.2025.110300","DOIUrl":"10.1016/j.compag.2025.110300","url":null,"abstract":"<div><div>In order to combine machine vision technology with single garlic clove extraction and garlic clove orientation adjustment devices in garlic planting mechanization to achieve single garlic clove orientational seeding, a real-time detection system of garlic clove bud for garlic clove orientation metering device was developed. Through the analysis of garlic clove images collected from the garlic clove orientation metering device, the curvature characteristics of the bud and root portions in the garlic cloves outer contour and their relationship were discovered, which could be used to identify the garlic clove bud. By sequentially used the methods of resampling the garlic clove outer contour curve after fitting, calculating the curvature of data points using consecutive data points with intervals, and selecting data points at uniform intervals on the specified garlic clove’s outer contour curve segment, it were eliminated gradually and effectively that interference of abnormal protrusions that may appear on the outer contour of the garlic clove, and highlighted the inherent curvature characteristics of the bud and the root. So the effective peak values of the curvature were extracted. The maximum, second-largest, and third-largest effective peak values were used as curvature characteristic parameters of the bud portion and root portion. And based on their proportional relationships, the position of the garlic clove bud was determined. The hardware system with an industrial control computer as the image processing unit was designed and three key parameters within the detection algorithm were optimized through experiments. They are the number of points to be sampled after curve fitting, the number of points to be interpolated when calculating the curvature of data points, and the number of points to be filtered when selecting data points at uniform intervals on the specified garlic clove outer contour curve segment. The optimal parameter combination obtained was 150, 6, and 4, respectively. Verification tests using the garlic clove orientation metering device showed an average detection accuracy of 96.22 %, and an average detection time of 114.9 ms per garlic clove image, which meets the real-time detection requirements of the garlic clove orientation metering device.</div></div>","PeriodicalId":50627,"journal":{"name":"Computers and Electronics in Agriculture","volume":"234 ","pages":"Article 110300"},"PeriodicalIF":7.7,"publicationDate":"2025-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143746583","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Optimizing the MoSt GG model a sensitivity-driven calibration for better grass growth forecasting","authors":"L. Bonnard ,&nbsp;L. Delaby ,&nbsp;M. Murphy ,&nbsp;E. Ruelle","doi":"10.1016/j.compag.2025.110288","DOIUrl":"10.1016/j.compag.2025.110288","url":null,"abstract":"<div><div>Grasslands offer an efficient and eco-friendly way to produce high-quality feed for ruminants, benefiting both livestock production and human nutrition. However, its high sensitivity to its environment makes its management challenging for farmers. Predicting week ahead grass growth results in better-informed decision making on farms. The Moorepark St Gilles Grass Growth Model (MoSt GG) has been used since 2018 to predict weekly grass growth on grassland farms across Ireland with 84 farms involved in 2023. The repeated use of the model on these farms has identified a need to improve its accuracy, which has been addressed in this study. First, a sensitivity analysis using the Morris method was conducted to identify the parameters that have the most influence on the model’s grass growth output, both on an annual and monthly time step. From that analysis, ten parameters were selected, all of which related either to temperatures, day length or nitrogen demand and availability for the grass. These ten parameters were calibrated using a semi-automatic iterative method of calibration on a dataset of 14 commercial farms containing four years of grass measurements. Nine iterations were necessary to calibrate the model resulting in a reduction of MAPE from 30.0% to 19.8% in its final calibrated version, and notably increasing the final R² from 0.58 to 0.71. Finally, the model was evaluated over a new dataset of ten commercial farms for four years. The evaluation confirmed the improvement of the model with a final MAPE of 19.1% and a R² of 0.67 compared to 30.1% and 0.57 respectively before the calibration. The calibration process of the MoSt GG model has significantly improved the model accuracy to predict on farm grass growth. This improvement is expected to be particularly valuable for farmers in their decision making process, providing them with more reliable on farm grass growth predictions.</div></div>","PeriodicalId":50627,"journal":{"name":"Computers and Electronics in Agriculture","volume":"235 ","pages":"Article 110288"},"PeriodicalIF":7.7,"publicationDate":"2025-03-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143724205","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Efficient detection of corn straw coverage in complex agricultural scenarios","authors":"Feiyun Wang ,&nbsp;Chengxu Lv ,&nbsp;Hanlu Jiang ,&nbsp;Yuxuan Pan ,&nbsp;Pengfei Guo ,&nbsp;Fupeng Li ,&nbsp;Liming Zhou","doi":"10.1016/j.compag.2025.110338","DOIUrl":"10.1016/j.compag.2025.110338","url":null,"abstract":"<div><div>Straw coverage serves as a critical indicator in the realm of conservation tillage. This study aims to fulfill the detection needs for straw coverage on edge monitoring platforms by initially capturing straw images through an onboard terminal and subsequently creating a dataset via data augmentation. We opted for SegNext as the foundational model and incorporated ResNet101 as the backbone to enhance the extraction of features specific to straw. To achieve a lightweight model without sacrificing detection accuracy, ResNet101 was utilized as the teacher model to mentor ResNet18 as the student model, with the training outcomes quantified using QAT. In tests conducted under multifactorial field scenarios, the QSR101-18 model achieved mIoU of 85.78 %, mAP of 95.98 % and Kappa of 86.25 %, surpassing SegNext by 1.44 %, 1.57 % and 1.32 %, respectively. The QSR101-18 model FLOPs and Params are 0.71G and 0.45 M respectively, which is about 1/27 and 1/100 of SegNext. When deployed on edge platforms and analyzed across varying straw coverage rates, QSR101-18 demonstrated an overall error of only 1.3 %, well within acceptable limits. The inference speed for a single image was just 16.32 ms, meeting the speed requirements for field operations. Consequently, the proposed QSR101-18 model demonstrates several key advantages, including a lightweight architecture, minimal error rates, robustness, and high accuracy. It effectively addresses the challenges posed by unstructured, fragmented straw and various environmental factors in detecting straw coverage, all while adhering to the speed constraints required for field operations on edge monitoring platforms.</div></div>","PeriodicalId":50627,"journal":{"name":"Computers and Electronics in Agriculture","volume":"235 ","pages":"Article 110338"},"PeriodicalIF":7.7,"publicationDate":"2025-03-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143724204","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Multi-step optimization design of pressure regulator for lateral inlet based on stepwise design of spring and structural","authors":"Xiaoran Wang ,&nbsp;Chen Zhang ,&nbsp;Guangyong Li","doi":"10.1016/j.compag.2025.110333","DOIUrl":"10.1016/j.compag.2025.110333","url":null,"abstract":"<div><div>Installing a pressure regulator for laterals (PRL) at the non-pressure compensated drip tape inlet offers a cost-effective, uniform pressure control solution for irrigation, especially in developing countries. PRL regulate both flow and pressure, requiring high performance. However, traditional optimization methods face challenges like extensive experimentation and the risk of compromising certain metrics while optimizing others. This study proposes a multi-step optimization method combining Computational Fluid Dynamics (CFD) and response surface experiments. Results show that with optimal spring parameters, PRLs achieve a pressure deviation (<em>α</em>) of under 5 %, an outlet pressure deviation from inlet pressure (<em>C</em>_V) under 10 %, and a pressure difference (Δ<em>P</em>) of less than 0.02 MPa across a 300–1000 L/h flow range. Unstable pressure at low flow is caused by a gap between the regulating cup and housing. Optimizing the outlet angle reduces pressure deviation from flow variations. Key factors influencing preset pressure (<em>P</em>_set) are spring stiffness (<em>K</em>) and pre-compression length (Δ<em>L</em>), followed by the bottom surface radius (<em>R</em>_bottom) and cup thickness (<em>R</em>_up). For <em>C</em>_V, <em>R</em>_bottom and <em>R</em>_up are most significant, with minimal impact from parameter interactions. For Δ<em>P</em>, <em>R</em>_bottom, <em>K</em>, Δ<em>L</em>, and <em>R</em>_up, with significant interactions, are key factors. Based on comprehensive evaluations, three PRL variants with preset pressures of 0.08, 0.10, and 0.12 MPa were developed, offering improved performance: Δ<em>H</em> under 0.05 MPa, ΔP under 0.012 MPa, <em>C</em>_V under 5 %, and <em>α</em> under 1.5 %. These optimized PRLs significantly outperform the original design and offer a broader range of products.</div></div>","PeriodicalId":50627,"journal":{"name":"Computers and Electronics in Agriculture","volume":"235 ","pages":"Article 110333"},"PeriodicalIF":7.7,"publicationDate":"2025-03-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143724203","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Remote and automated detection of Asian hornets (Vespa velutina nigrithorax) at an apiary, using spectral features of their hovering flight sounds","authors":"Harriet Hall ,&nbsp;Martin Bencsik ,&nbsp;Nuno Capela ,&nbsp;José Paulo Sousa ,&nbsp;Dirk C. de Graaf","doi":"10.1016/j.compag.2025.110307","DOIUrl":"10.1016/j.compag.2025.110307","url":null,"abstract":"<div><div>Asian hornets (<em>Vespa velutina nigrithorax</em>) are an invasive species that have spread across Europe since 2004. As <em>V.velutina</em> largely predate on honeybees, assessing their presence at apiaries would be useful for invasive species control programmes and beekeepers to help protect their hives. At present, hornet monitoring techniques are both costly and time consuming. A promising alternative is a remote detection strategy at apiaries, which would promote straightforward, non-invasive data acquisition. The remote capture of flight acoustics should benefit hornet detection as wingbeat frequencies have previously been described as ‘the fingerprint’ of some flying invertebrate species. We here demonstrate a non-invasive method of <em>V.velutina</em> detection using their hovering flight sounds, captured by microphones that can be left at an apiary over the long-term. Paired with a training algorithm (principal component analysis and discriminant function analysis) that discriminates between hornet flight and other external noises (honeybee flight sounds and general background noise), we demonstrate that hornet hovering acoustics exhibit specific spectral features that promote the detection of individuals at an apiary. The training algorithm in our study was highly accurate (98.7 %) when testing just under 1-hour of apiary recordings. Utilising two-dimensional-Fourier-transforms has also benefited this algorithm, as the analysis technique is ideal for identifying repeating features in sound/vibrational data, which are an inherent consequence of hovering hornet sounds. The experimental design and training algorithm used in this study have demonstrated excellent potential for hornet detection in the field, and are now ready to be tested on long-term, continuous data to further assess their success.</div></div>","PeriodicalId":50627,"journal":{"name":"Computers and Electronics in Agriculture","volume":"235 ","pages":"Article 110307"},"PeriodicalIF":7.7,"publicationDate":"2025-03-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143724541","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}