Computers and Electronics in Agriculture最新文献

{"title":"Developing an IoT-driven delta robot to stimulate the growth of mulberry branch cuttings cultivated aeroponically using machine vision technology","authors":"Osama Elsherbiny ,&nbsp;Jianmin Gao ,&nbsp;Ming Ma ,&nbsp;Waqar Ahmed Qureshi ,&nbsp;Abdallah H. Mosha","doi":"10.1016/j.compag.2025.110111","DOIUrl":"10.1016/j.compag.2025.110111","url":null,"abstract":"<div><div>Machine vision plays a pivotal role in automatically monitoring the growth of mulberry branch cuttings (Mbc) in aeroponic systems, ensuring productivity, quality, and sustainability. However, the challenge lies in the varying bud growth rates, with some taking longer to break dormancy, leading to inconsistent development and delays in root formation. This paper aims to develop an Internet of Things (IoT)-integrated delta robot, equipped with advanced camera data acquisition and intelligent processing. It is intended to enhance aeroponic systems by precisely and rapidly detecting the Mbc growth state and applying growth stimulation. The system framework requires tiny machine learning models specifically designed to function efficiently on IoT hardware with limited power and resources, such as the lightweight versions of Tiny-YOLO, including YOLOv8-world, YOLOv9, YOLOv10, and YOLOv11. These models were trained on 3,000 images captured from three distinct camera perspectives—side view, elevated view, and angled view—during the growth of Mbc. Further optimization was achieved by progressively refining the weights through stepwise training. Artificial neural networks, for instance Back-Propagation Neural Networks (BPNN) and Elastic Net (ELNET), were deployed to compute the X, Y, and Z coordinates of the robot arm. Alongside, the look-up table approach efficiently identified the Mbc locations by referencing pre-stored data corresponding to the target coordinates. The experimental outcomes indicated that the optimized Tiny-YOLOv9 (Pr = 98.3 %, Re = 98.5 %, Fm = 98.4 %, mAP_50–90 = 87.3 %) outperformed other models in both classification and localization of mulberry branches. Data fusion from three cameras with BPNN and ELNET models substantially outshined the use of a single camera. Moreover, the BPNN models for the arm axes (X, Y, Z) exhibited superior accuracy compared to ELNET. The BPNN recorded R² values of 0.999 for all three axes (X, Y, and Z), with corresponding RMSE values of 0.005 (MAE = 0.004), 0.006 (MAE = 0.005), and 0.013 (MAE = 0.010), respectively. This work can assist the agricultural community in monitoring plant growth, enabling timely and effective management decisions. It also holds great promise for expanding our methodology to include other crops in future aeroponic systems.</div></div>","PeriodicalId":50627,"journal":{"name":"Computers and Electronics in Agriculture","volume":"232 ","pages":"Article 110111"},"PeriodicalIF":7.7,"publicationDate":"2025-02-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143388042","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":"Forest biomass carbon stock estimates via a novel approach: K-nearest neighbor-based weighted least squares multiple birth support vector regression coupled with whale optimization algorithm","authors":"Niannian Deng ,&nbsp;Renpeng Xu ,&nbsp;Ying Zhang ,&nbsp;Haoting Wang ,&nbsp;Chen Chen ,&nbsp;Huiru Wang","doi":"10.1016/j.compag.2025.110020","DOIUrl":"10.1016/j.compag.2025.110020","url":null,"abstract":"<div><div>Multiple birth support vector regression (MBSVR) provides fast computation and superior performance but overlooks local sample information and has challenges in parameter selection. The traditional least squares models boast fast computational speed but lack robustness and may struggle with noise and outliers. The carbon storage estimates are easily affected by noise and interference points. MBSVR and least squares models are only partially effective in carbon storage estimates. Consequently, we propose least squares multiple birth support vector regression (LSMBSVR) and K-nearest neighbor-based (KNN) weighted least squares multiple birth support vector regression (WLSMBSVR), which have the following merits. Firstly, both models inherit the strengths of MBSVR. Secondly, they exhibit enhanced fitting accuracy, robust stability, and remarkable anti-interference capability. Thirdly, LSMBSVR offers a faster training speed and maintains a comparable regression performance to MBSVR. Fourthly, WLSMBSVR considers the local information, enhancing its anti-interference capability. Lastly, we employ the whale optimization algorithm (WOA) to improve the effectiveness of parameter selection. Experiment results indicate that our models can be more effective on carbon storage, synthetic, and UCI datasets than compared models, verifying the broad application value of our models.</div></div>","PeriodicalId":50627,"journal":{"name":"Computers and Electronics in Agriculture","volume":"232 ","pages":"Article 110020"},"PeriodicalIF":7.7,"publicationDate":"2025-02-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143379118","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":"Agricultural data privacy and federated learning: A review of challenges and opportunities","authors":"Rahool Dembani ,&nbsp;Ioannis Karvelas ,&nbsp;Nur Arifin Akbar ,&nbsp;Stamatia Rizou ,&nbsp;Domenico Tegolo ,&nbsp;Spyros Fountas","doi":"10.1016/j.compag.2025.110048","DOIUrl":"10.1016/j.compag.2025.110048","url":null,"abstract":"<div><div>The rapid digitalization of agriculture has resulted in an unprecedented surge in data collection, necessitating this way the privacy protection in innovative data analytics solutions. Federated Learning emerges as a promising solution since it allows for collaborative model training across decentralized data sources without sharing raw data. This review explores the use of Federated Learning in agriculture, focusing on privacy-preserving methods. We thoroughly reviewed a large corpus of relevant research, examining several Federated Learning types and their application to agricultural scenarios, such as pest and disease detection, crop yield prediction, and resource management. Our findings underscore Federated Learning’s potential to revolutionize privacy-preserving data analysis in agriculture by enabling better decision-making through aggregated insights from various farms, while retaining data confidentiality. At the same time, a number of technical complications arise, including data heterogeneity, communication impediments, and limited computational capabilities in rural areas. Data ownership, fairness, and stakeholder trust are significant barriers to widespread use in practice. The present study provides research gaps that need to be addressed to fully use the potential of Federated Learning in agriculture. Tailoring the design of Federated Learning algorithms and adhering to the nature of agricultural data and its peculiarities can promote the enhancement of agriculture-friendly frameworks to ensure privacy-preserving mechanisms for agriculture-oriented applications, and the development of frameworks that bear ethical issues in mind and facilitate farmers-based equitable benefit distribution. Since Federated Learning can potentially change the landscape of data-driven agriculture by allowing collaborative data analytics without compromising privacy, it is highly important to overcome the technological and ethical barriers demonstrated in this study, maximizing its impact on<!--> <!-->sustainable farming practices and innovations.</div></div>","PeriodicalId":50627,"journal":{"name":"Computers and Electronics in Agriculture","volume":"232 ","pages":"Article 110048"},"PeriodicalIF":7.7,"publicationDate":"2025-02-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143376454","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":"A methodology for the realistic assessment of 3D point clouds of fruit trees in full 3D context","authors":"Bernat Lavaquiol-Colell ,&nbsp;Alexandre Escolà ,&nbsp;Ricardo Sanz-Cortiella ,&nbsp;Jaume Arnó ,&nbsp;Jordi Gené-Mola ,&nbsp;Eduard Gregorio ,&nbsp;Joan R. Rosell-Polo ,&nbsp;Jérôme Ninot ,&nbsp;Jordi Llorens-Calveras","doi":"10.1016/j.compag.2025.110082","DOIUrl":"10.1016/j.compag.2025.110082","url":null,"abstract":"&lt;div&gt;&lt;div&gt;The aim of this paper is to address the lack of standard methodologies for the assessment of 3D point clouds. We present a methodology to realistically assess the accuracy of 3D point clouds, enabling the evaluation in a full 3D context rather than based on isolated points. Additionally, it introduces three significant innovations: a) it bridges the gap related to the unknown error of the reference ground-truth point cloud; b) it provides separate metrics for location error and reconstruction error; and c) it introduces a procedure to compute the location error that eliminates the bias in the selection of point-pair picking between the DGT points and their corresponding pairs in the point cloud being assessed.&lt;/div&gt;&lt;div&gt;The geometry and structure of trees are related to the vegetative parameters and productivity in fruit orchards. In consequence, obtaining a precise and accurate geometric characterization of canopies is of interest for implementing site-specific management strategies that optimize input rates and minimize the costs and environmental risks of agricultural operations. Among the different sensing technologies, sensors based on the principle of light detection and ranging (LiDAR) have emerged as the primary choice for accurate geometric characterization of orchards. However, to make informed orchard management decisions based on LiDAR-derived geometric and structural data, it is essential to assess the accuracy of LiDAR-based scanning systems. Unfortunately, there is currently a lack of standard methodologies to evaluate the accuracy of LiDAR-based systems in agricultural environments. This research paper presents a novel methodology to assess the location error and the reconstruction error of 3D point clouds in full 3D context. The methodology involves comparing LiDAR-derived point clouds to an accurate high-resolution 3D digital ground truth (DGT) obtained using digital photogrammetric techniques. One of the main difficulties when using a reference point cloud to assess point cloud errors is the selection of the points to be compared so that they can be considered as corresponding point pairs. When developing the methodology, four procedures of point pair selection and distance calculation were compared. The best performing procedure was selected and proposed as a standard for accuracy assessment of 3D point clouds. The proposed procedure minimizes the error attributed to the selection of the corresponding point pairs between the assessed point cloud and the reference DGT point cloud. Subsequently, the proposed methodology was tested and validated by assessing the accuracy of 46 different point clouds.&lt;/div&gt;&lt;div&gt;The conclusions regarding the accuracy, applicability, and practical utility of the proposed methodology are supported by the determination of reconstruction errors and location errors in 46 point clouds obtained with the 3 different MTLS systems operated with different settings. The proposed methodology will be v","PeriodicalId":50627,"journal":{"name":"Computers and Electronics in Agriculture","volume":"232 ","pages":"Article 110082"},"PeriodicalIF":7.7,"publicationDate":"2025-02-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143372365","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":"Enhanced fish stress classification using a cross-modal sensing fusion system with residual depth-separable convolutional networks","authors":"Wentao Huang ,&nbsp;Yunpeng Wang ,&nbsp;Wenhao He ,&nbsp;Xiaoshuan Zhang","doi":"10.1016/j.compag.2025.110038","DOIUrl":"10.1016/j.compag.2025.110038","url":null,"abstract":"<div><div>Fish stress classification is crucial for intelligent and precision aquaculture. However, single-modal information often leads to inaccurate classification results due to incomplete feature representation. To address this, this study proposes a cross-modal system that integrates RGB, infrared (IR), and impedance (IMP) sensor data to improve the classification performance of fish stress states. A unique dataset was created through experiments conducted in a controlled environment, encompassing various levels of fish stress features. Subsequently, a stress classification model was developed using multi-modal fusion methods, with RDC-Net as the core network. The network adopts early, mid, and late fusion strategies to accommodate the characteristics of the data. Specifically, the model introduces a cross-residual fusion module, inspired by the residual blocks in ResNet, to implement a multi-level data fusion approach. Results indicate that the model achieves an accuracy of 94.02%, a recall of 99.64%, a precision of 94.33%, and an F1-score of 0.9690 on the test set. The model outperforms other ablative methods in terms of accuracy. Feature extraction experiments show that RDC-Net can effectively extract and identify relevant features in the model. Additionally, a real-time fish stress grading application based on RDC-Net is proposed, providing a feasible solution for fish stress classification strategies.</div></div>","PeriodicalId":50627,"journal":{"name":"Computers and Electronics in Agriculture","volume":"231 ","pages":""},"PeriodicalIF":7.7,"publicationDate":"2025-02-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143349808","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":"A parallel dual-arm robotic control method of white asparagus based on moving-looking-harvesting coordination and asynchronous harvest cooperation","authors":"Ping Zhang ,&nbsp;Nianzu Dai ,&nbsp;Zinuo Wang ,&nbsp;Jin Yuan ,&nbsp;Zhenbo Xin ,&nbsp;Xuemei Liu ,&nbsp;Georgios Papadakis","doi":"10.1016/j.compag.2025.110046","DOIUrl":"10.1016/j.compag.2025.110046","url":null,"abstract":"<div><div>For an efficient, low-damage selective harvesting robot for white asparagus, a parallel dual-arm control method based on moving-looking-harvesting coordination and asynchronous spears harvesting cooperation is proposed. This approach aims to enable continuous, non-stop harvesting along the ridge while minimizing damage to the spears. Firstly, a parallel harvesting mode with independent harvesting areas for the dual arms was designed to reduce collisions between the robotic arms and simplify control complexity. Secondly, an efficient dual-arm cooperative harvesting algorithm, aimed at load balancing for multiple randomly distributed asparagus, was proposed. Then, a coordinated moving-looking-harvesting control strategy was developed to synchronize the robot’s movement, asparagus identification, and the harvesting operation using the two end-effectors. Finally, a prototype of the selective harvesting system was constructed, and its performance was evaluated in the field. The simulation analysis of the cooperative harvesting algorithm indicated that the shortest-time-based first-see-harvest (ST-FSH) path planning strategy outperformed two alternative methods. The dual-arm harvesting saved 45.17 % of the time and increased the harvest success rate by 3.19 % compared with the single-arm harvesting, while maintaining workload balance. Field trials demonstrated an asparagus recognition rate of 82.6 %, with an average detection time of 33 ms, a successful harvest rate of 92.3 % for recognized asparagus, and average robotic arm movement time and end-effector harvest time of 1.7 s and 5.7 s, respectively. The system achieved an asparagus damage rate of 7.2 %. The results confirm the feasibility of the proposed efficient, low-damage harvesting strategy, providing a solid foundation for the development of selective harvesting robots for white asparagus.</div></div>","PeriodicalId":50627,"journal":{"name":"Computers and Electronics in Agriculture","volume":"232 ","pages":"Article 110046"},"PeriodicalIF":7.7,"publicationDate":"2025-02-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143349019","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":"Research on predictive control of a novel electric cleaning system for combine harvester based on data-driven","authors":"Zhihao Zhu ,&nbsp;Xiaoyu Chai ,&nbsp;Lizhang Xu ,&nbsp;Li Quan ,&nbsp;Chaochun Yuan ,&nbsp;Shuofeng Weng ,&nbsp;Guangqiao Cao ,&nbsp;Weijun Jiang","doi":"10.1016/j.compag.2025.110075","DOIUrl":"10.1016/j.compag.2025.110075","url":null,"abstract":"<div><div>In order to optimize the efficiency of the combine harvester cleaning system, this research introduces a data-driven control approach merging subspace model identification and event-triggered adaptive model predictive control (ETAMPC) under a linear parameter-varying (LPV) system structure. This method addresses the challenges of modeling and controlling the cleaning system, treating it as a “black box”. It is further applied to a newly designed electric cleaning system (ECS), solving the problem of difficult real-time adjustment of fan speed and vibrating sieve frequency caused by mechanical coupling in conventional cleaning systems, and achieved coordinated control over fan speed and vibrating sieve frequency to reduce loss rate and impurity rate. The simulation results show that the accuracy of the predicted output of the constructed ECS identification model exceeds 85%, and the designed ETAMPC strategy not only exhibits good effect of performance tracking (with tracking errors remaining below 10% under random disturbances) but also effectively reduce computational load by approximately 50%. Field tests indicate that the designed ECS can reduce cleaning losses by 16% to 19% and impurities by 13% to 27%. This system offers a new pathway to enhance the operating performance of combine harvesters.</div></div>","PeriodicalId":50627,"journal":{"name":"Computers and Electronics in Agriculture","volume":"232 ","pages":"Article 110075"},"PeriodicalIF":7.7,"publicationDate":"2025-02-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143372364","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":"Moisture regain sensing method for seed cotton under multi-factor fusion","authors":"Mianzhe Hong ,&nbsp;Liang Fang ,&nbsp;Huting Wang ,&nbsp;Hongwei Duan ,&nbsp;Jinqiang Chang ,&nbsp;Hao Li ,&nbsp;Ruoyu Zhang","doi":"10.1016/j.compag.2025.110073","DOIUrl":"10.1016/j.compag.2025.110073","url":null,"abstract":"<div><div>To enhance the accuracy of moisture regain (MR) sensing in seed cotton during mechanical harvesting, a capacitance, resistance, density and temperature (CRDT) fusion method for seed cotton MR sensing was proposed in this study. Initially, seed cotton samples exhibiting various MR levels were prepared using a saturated salt solution method to control the environmental humidity. Subsequently, an experimental platform was designed to simulate the changing temperature and density in the packaging room of a cotton picker. The capacitance and resistance of the “Zhongmian 113” seed cotton samples were measured under eight temperature gradients (5, 10, 15, 20, 25, 30, 35, and 40 °C) and eleven density gradients (149.28, 155.50, 162.26, 169.64, 177.71, 186.60, 196.42, 207.33, 219.53, 233.25 and 248.80 kg/m³). The Pearson correlation analysis showed that MR had a high correlation with capacitance (0.82) and resistance (−0.88). Capacitance displayed a linear relationship with both density (R² = 0.99) and temperature (R² = 0.93), whereas resistance demonstrated a linear relationship with density (R² = 0.94) and an exponential relationship with temperature (R² = 0.97). Furthermore, multiple linear regression (MLR), back propagation neural network (BPNN), support vector machine (SVM) and random forest (RF) were used to build models for seed cotton MR sensing under varying conditions. The results showed that the CRDT fusion model outperformed the models constructed based solely on capacitance or resistance. Among the modeling strategies, the BPNN exhibited the best performance, with an R² of 0.99 and an RMSE of 0.21 %. This study will provide a valuable reference for the development of seed cotton MR sensor in cotton mechanical harvesting.</div></div>","PeriodicalId":50627,"journal":{"name":"Computers and Electronics in Agriculture","volume":"232 ","pages":"Article 110073"},"PeriodicalIF":7.7,"publicationDate":"2025-02-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143350506","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":"CattlePartNet: An identification approach for key region of body size and its application on body measurement of beef cattle","authors":"Zixia Hou ,&nbsp;Qi Zhang ,&nbsp;Bin Zhang ,&nbsp;Hongming Zhang ,&nbsp;Lyuwen Huang ,&nbsp;Meili Wang","doi":"10.1016/j.compag.2025.110013","DOIUrl":"10.1016/j.compag.2025.110013","url":null,"abstract":"<div><div>Addressing the challenges associated with manual measurement of body sizes for beef cattle, the potential for inducing stress responses in animals, and the inefficiencies inherent in such labor-intensive tasks, a novel three-dimensional (3D) point-based deep learning (DL) network named CattlePartNet is proposed. This innovative network is redesigned to segment point cloud data (PCD) of cattle into crucial body regions, thereby facilitating and optimizing the measurement process. The newly proposed CattlePartNet adopts the base network of PointNet++ as its backbone network, where the parallelized patch-aware attention module and depth-wise separable convolutions are freshly incorporated to lower the risk of overfitting. Additionally, it incorporates the Sophia optimizer, a novel and highly efficient optimization algorithm, instead of the Adam optimizer in PointNet++. Impressively, CattlePartNet outperforms the PointNet++ backbone network with a 2.4 % improvement in mean Intersection over Union (mIoU), as demonstrated on the ShapeNetPart dataset, which is widely regarded as an established benchmark for part segmentation tasks within the domain of PCD analysis. Leveraging the established network, CattlePartNet is trained on a dataset of cattle PCD for automated segmentation of key body regions, achieving an impressive mIoU of 91.7 %. The body sizes of cattle are categorized into two types: linear and curvilinear. For linear body sizes: body height (BH), body length (BL), and hip height (HH), measurement points are identified and extracted using sophisticated contour extraction techniques such as Alpha Shapes, mean curvature, and Gaussian curvature. Curvilinear body sizes: chest girth (CG) and abdominal circumference (AC) are measured through slice interception and cubic B-spline curve methods. The mean relative errors for the 5 body sizes—BH, BL, CG, AC, and HH—are reported as 4.96 %, 5.47 %, 6.04 %, 5.68 %, and 5.49 %, respectively. In comparison to traditional measurement of beef cattle, CattlePartNet precisely extracts key regions of body size from PCD, demonstrating robustness in generalizing segmentation across diverse cattle breeds and showing potential for segmenting PCD of other large livestock species. Furthermore, the measurement algorithm of 5 body sizes accurately localizes key measurement points within each region, providing essential support for breeding applications such as health assessment and production performance measurement.</div></div>","PeriodicalId":50627,"journal":{"name":"Computers and Electronics in Agriculture","volume":"232 ","pages":"Article 110013"},"PeriodicalIF":7.7,"publicationDate":"2025-02-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143349018","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":"Insect-YOLO: A new method of crop insect detection","authors":"Nan Wang ,&nbsp;Shaowen Fu ,&nbsp;Qiong Rao ,&nbsp;Guiyou Zhang ,&nbsp;Mingquan Ding","doi":"10.1016/j.compag.2025.110085","DOIUrl":"10.1016/j.compag.2025.110085","url":null,"abstract":"<div><div>The utilization of the pest monitoring and reporting system has gained widespread adoption for automating the surveillance of field-dwelling pests, serving as a viable alternative to the labor-intensive and time-consuming manual inspection methods. Nevertheless, the heterogeneous spectrum and variable sizes of crop pests, coupled with the imperative to manage costs associated with camera lenses employed in practical agricultural scenarios, result in low-resolution images. This low resolution significantly amplifies the intricacy of pest identification. Our research is dedicated to the detection of insects in low-resolution images, we collected a large dataset of low-resolution images of common pests from agricultural fields, with resolutions ranging from 8 to 12 million pixels, and deployed the Insect-YOLO model based on this dataset. Tailored for capturing pests on diverse crops, Insect-YOLO boasts streamlined parameters, swift detection speeds, and exceptional accuracy. Enhanced by the Convolutional Block Attention Module (CBAM), it systematically extracts complex pest features, integrating multi-scale information to optimize feature representation. In comparative evaluations against YOLO v5, v7, v8, RetinaNet, and Faster R-CNN, Insect-YOLO demonstrated exceptional performance, achieving a mean Average Precision at IoU 0.5 (mAP₅₀) of 93.8%, highlighting its superiority in pest detection. Simultaneously, linear regression analysis was performed to assess the correlation between the computer-detected and manually counted insect numbers, revealing a strong correlation that underscores the efficacy of our method. Ultimately, the pest detection algorithm was integrated into the “Remote Pest Monitoring and Analysis System” of the Agricultural IoT Monitoring Platform. This integration enables high accuracy and efficiency in detecting diverse pests from real-time, low-resolution field images and constitutes a critical component of a comprehensive pest monitoring system, serving as a foundation for pest prediction and intelligent monitoring technologies.</div></div>","PeriodicalId":50627,"journal":{"name":"Computers and Electronics in Agriculture","volume":"232 ","pages":"Article 110085"},"PeriodicalIF":7.7,"publicationDate":"2025-02-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143350104","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}