Computers and Electronics in Agriculture最新文献

{"title":"Optimized learning pipeline for predicting future dissolved oxygen levels in aquaculture using time-series data","authors":"Mingyan Wang ,&nbsp;Shahbaz Gul Hassan ,&nbsp;Ferdous Sohel ,&nbsp;Tonglai Liu ,&nbsp;Min He ,&nbsp;Xuekai Gao ,&nbsp;Caijin Xie ,&nbsp;Xiwen Deng ,&nbsp;Shuangyin Liu ,&nbsp;Longqin Xu","doi":"10.1016/j.compag.2025.110509","DOIUrl":"10.1016/j.compag.2025.110509","url":null,"abstract":"<div><div>To achieve long-term accurate predictions of dissolved oxygen in aquaculture environments. In this paper, we propose the Time varying filter based empirical mode decomposition (TVFEMD) - Permutation Entropy (PE) -Temporal Convolutional Networks (TCN) - bidirectional gated recurrent units (BiGRU)-Improved Sparrow Search Algorithm (ISSA) (TPBI) model. First, dissolved oxygen data are applied to the Time-Variant Filtered Empirical Mode Decomposition (TVFEMD) model to remove noise factors in the data. Permutation Entropy (PE) is then applied to reconstruct the data and reduce its complexity. Additionally, Temporal Convolutional Networks (TCN) and Bidirectional Gated Recurrent Units (BiGRU) are combined to extract features from the denoised data, improving the model’s learning efficiency and prediction accuracy. Based on this, the Dynamic Opposite Learning Strategy Improved Sparrow Search Algorithm (ISSA) is introduced to optimize hyperparameters such as batch size and the number of hidden layer units.</div><div>The framework was applied to predict dissolved oxygen data from an aquaculture farm in Guangdong Province, with future 1-step, 3-step, and 6-step prediction experiments. The experimental results show that the proposed model outperforms the comparison models in predicting dissolved oxygen, particularly excelling in long-term predictions (6 steps). In terms of Mean Absolute Error (MAE), compared to models such as RNN, BiGRU, CEEMDAN-PE-TCN-BiGRU-ISSA, and TVFEMD-PE-TCN-BiGRU-SSA, the proposed model improved dissolved oxygen prediction by 50%, 55.2%, 50%, and 27.7%, respectively.</div><div>Ablation experiments were conducted to verify the effectiveness of all components. In terms of Mean Absolute Error (MAE) and Root Mean Square Error (RMSE), dissolved oxygen prediction improved by 41.3% and 21.0%, respectively. The outstanding performance of this framework in long-term predictions of dissolved oxygen provides effective support for precise environmental control and early warning in aquaculture.</div></div>","PeriodicalId":50627,"journal":{"name":"Computers and Electronics in Agriculture","volume":"236 ","pages":"Article 110509"},"PeriodicalIF":7.7,"publicationDate":"2025-05-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143941571","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":"Shifting strategy for power shift tractors based on digital Twin-Driven reinforcement learning","authors":"Chang Feilong,&nbsp;Lu Zhixiong,&nbsp;Deng Xiaoting","doi":"10.1016/j.compag.2025.110513","DOIUrl":"10.1016/j.compag.2025.110513","url":null,"abstract":"<div><div>To address the unstable power output and low operational efficiency in high-power power-shift tractors (PST) caused by engine performance variations and traction resistance fluctuations, this study proposes a power-shift strategy based on reinforcement learning and digital twin technology. A digital twin system is developed to achieve real-time synchronization between the physical PST and its virtual model through multi-source sensor data acquisition and standardized signal processing, enabling bidirectional interaction and dynamic environment simulation. The proposed strategy integrates a twin-delayed deep deterministic policy gradient (TD3) reinforcement learning framework to mitigate Q-value overestimation and enable adaptive optimization of shifting decisions under complex operating conditions. Compared with traditional optimization methods such as dynamic programming and conventional neural networks, the TD3-based approach demonstrates superior adaptability and control stability, particularly in maintaining smooth shifting and continuous power delivery under varying load conditions. Furthermore, to address throttle fluctuation during gear transitions, a fuzzy PID throttle controller is introduced, which dynamically adjusts PID gains based on real-time throttle deviation and its rate of change. Experimental results show that the proposed method significantly reduces vehicle speed tracking errors and fuel consumption while improving gear-shift smoothness. Specifically, the mean engine torque and fuel consumption tracking errors remain below 6.11 N·m and 1.86 g·(kW·h)^–1, respectively. Compared to traditional strategies, the method achieves a lower mean speed tracking error (0.0121 m·s^–1), fuel consumption rate (231.21 g·(kW·h) ^–1), and total number of shifts (39).This study presents an effective and intelligent gear-shifting solution for PSTs and offers valuable insights for the broader application of reinforcement learning in agricultural machinery control.</div></div>","PeriodicalId":50627,"journal":{"name":"Computers and Electronics in Agriculture","volume":"236 ","pages":"Article 110513"},"PeriodicalIF":7.7,"publicationDate":"2025-05-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143941572","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":"Deep learning assisted real-time nitrogen stress detection for variable rate fertilizer applicator in wheat crop","authors":"Narendra Singh Chandel ,&nbsp;Dilip Jat ,&nbsp;Subir Kumar Chakraborty ,&nbsp;Abhishek Upadhyay ,&nbsp;A. Subeesh ,&nbsp;Pooja Chouhan ,&nbsp;Monika Manjhi ,&nbsp;Kumkum Dubey","doi":"10.1016/j.compag.2025.110545","DOIUrl":"10.1016/j.compag.2025.110545","url":null,"abstract":"<div><div>An early and rapid detection of nitrogen (N) stress in field crops is crucial to mitigating nutrient deficiency and achieving sustainable crop yield. Although numerous methods and equipment have been developed to monitor crop N stress and fertilizer application thereof, many of these technologies face significant limitations in terms of costs, accuracy, integration, etc. This study reports the development of a Variable Rate fertilizer Application (VRA) system assisted by Deep Learning (DL) model deployed embedded system to enable rapid RGB image-based detection of nitrogen stress in wheat crop and subsequent application of N fertilizer. AlexNet DL model resulted in precision, recall, and F1-score as 0.977, 0.973, and 0.973, respectively; for classifying N stress into three classes. The developed VRA could operate in sync with embedded system at an operational speed of 0.4 m/s with a field capacity of 0.32 ha/h in a 26 DAS wheat crop. The effectivity of the VRA was evaluated by vegetation indices (ExG, RGRI, VARI and NGRDI) with drone assisted RGB images before and after VRA operation; there was a consistent difference in before and after average index values for ExG (0.2046 and 0.2917) and VARI (0.1478 and 0.2454). These results are indicative of the uniformity of operation by VRA throughout the field. The average percentage N fertilizer saving under VRA as compared to traditional technique was 37.53 % with an insignificant (p &lt; 0.05) difference in yield. This study delivers a real-time effective technique for precise classification of N stress and its real-time mechanized management in wheat crop.</div></div>","PeriodicalId":50627,"journal":{"name":"Computers and Electronics in Agriculture","volume":"237 ","pages":"Article 110545"},"PeriodicalIF":7.7,"publicationDate":"2025-05-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143946850","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 computational framework for modeling and predicting maize senescence: integrating UAV phenotyping, logistic growth, and genomics","authors":"Alper Adak ,&nbsp;Aaron J. DeSalvio ,&nbsp;Seth C. Murray","doi":"10.1016/j.compag.2025.110471","DOIUrl":"10.1016/j.compag.2025.110471","url":null,"abstract":"<div><div>Advances in sensing technologies have led to the emergence of a new paradigm in plant biology and predictive plant breeding able to improve the precision in quantifying and predicting physiological traits such as plant senescence. Here, 517 recombinant inbred lines (RILs) were previously genotyped and phenotyped in this study using an unoccupied aerial system (UAS also known as UAV or drone) equipped with an RGB sensor, enabling efficient monitoring of senescence at multiple developmental stages across 14 flights from 28 to 128 days after planting. Temporal senescence was scored in the last five flights and uniquely subjected to a logistic growth model (R² = 0.99 ± 0.01). Days to Senescence (DTSE) and a modified Grain Filling Period (GFP) were introduced in this study, derived from the logistic growth model using temporal senescence data.</div><div>The study also explored the predictive power of logistic growth model-driven genomic (M1) and combined genomic and phenomic (M2) models. The combined model (M2), incorporating phenomic data from vegetation indices (NGRDI and ExR) collected before senescence, outperformed the genomic model (M1), particularly in challenging scenarios involving untested RILs and time points (CV2: 0.32 vs 0.48; CV00: 0.22 vs 0.33), showcasing potential for predictive breeding of delayed senescence. M1 achieved prediction abilities of 0.45 and 0.38 for DTSE and GFP, respectively, which improved to 0.47 and 0.40 with M2.</div><div>Overall, this research advances the prediction of temporal senescence dynamics through computational frameworks integrating phenotyping, modeling, and genomic data. These advancements enable the selection of genotypes with optimized senescence rate.</div></div>","PeriodicalId":50627,"journal":{"name":"Computers and Electronics in Agriculture","volume":"237 ","pages":"Article 110471"},"PeriodicalIF":7.7,"publicationDate":"2025-05-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143936320","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":"Forecasting the mid-and-long-term crop evapotranspiration for winter wheat in China using combination models based on public weather forecast data","authors":"Liqin Gong ,&nbsp;Zhigong Peng ,&nbsp;Baozhong Zhang ,&nbsp;Zheng Wei ,&nbsp;Guiyu Yang ,&nbsp;Jiabing Cai ,&nbsp;Xiling Zhang ,&nbsp;Yingduo Yu","doi":"10.1016/j.compag.2025.110504","DOIUrl":"10.1016/j.compag.2025.110504","url":null,"abstract":"<div><div>Forecasting the mid-and-long-term crop evapotranspiration (ET_c) for winter wheat in China is beneficial for crop water dynamic management and can significantly improve water use efficiency. This research has important practical implications for alleviating the mismatch of timing between the crop water requirement and precipitation during the growth period of winter wheat. Based on the principle of minimizing the root mean square error, the combination models for 1–30 day forecast periods at 2,391 stations were established using six models, namely, Hargreaves (HG), McClound (MC), Makkink (MK), Temperature Penman–Monteith (PMT), Priestley–Taylor (PT), and Penman–Monteith Forecast (PMF), as the preferred models for estimating reference crop evapotranspiration (ET_o).The crop coefficient approach is used to forecast the ET_c for each month during the growth period of winter wheat. Results show the following: (1) The distributed parameters of HG model, MC model, MK model, and PT model were calibrated using the least squares method based on the availability of collected meteorological data in the past 50 years. (2) The forecast models are ranked in descending order of ET_o forecast accuracy. The HG and PMF models are favored for short- and medium-term forecasts, while the MC, PT, and PMF models are preferred for long-term forecasts. At the spatial scale, each forecast model exhibits distinct spatial distribution characteristics. (3) Based on the mid-and-long-term reference evapotranspiration forecast using the combination models with an average monthly error of less than 11 %. The spatiotemporal variations in the forecasted ET_c for winter wheat are consistent with the research of Sun et al. (2013). The results of this research provide valuable references for determining ET_o forecast model parameters and selecting appropriate forecast models at different spatial and temporal scales, which reduces the impact of weather forecast uncertainty. This understanding offers theoretical and practical support for mid-and-long-term ET_c forecasting.</div></div>","PeriodicalId":50627,"journal":{"name":"Computers and Electronics in Agriculture","volume":"236 ","pages":"Article 110504"},"PeriodicalIF":7.7,"publicationDate":"2025-05-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143941453","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":"PhenoBee: Drone-based robot for advanced field in vivo contact-based phenotyping in agriculture","authors":"Ziling Chen ,&nbsp;Xuan Li ,&nbsp;Tianzhang Zhao ,&nbsp;Xing Wei ,&nbsp;Zhihang Song ,&nbsp;Jialei Wang ,&nbsp;Jin Jian","doi":"10.1016/j.compag.2025.110483","DOIUrl":"10.1016/j.compag.2025.110483","url":null,"abstract":"<div><div>Spectral imaging has been widely applied for soybean phenotyping to find and maintain favorable traits. Specifically, in soybean phenotyping, hyperspectral imaging through contact-based proximal sensing demonstrates better signal-to-noise ratio and resolution compared to remote sensing. However, it has not been adapted for large-scale field applications due to its low throughput and high labor costs. Additionally, no automation solution has been developed to collect in vivo contact-based hyperspectral images of soybean plants. In this study, a novel drone-based robotic system was developed to automate the collection of in vivo contact-based hyperspectral images in the field. The system consists of a machine vision system to detect and estimate the pose of soybean leaflets, an articulated robotic arm with specialized control and path planning algorithms to operate contact-based sensors to grasp and image the leaf, and a customized high-payload drone to provide mobility for sampling at different locations across a field. The average accuracy of the optimized machine vision algorithm is 95.88% for leaf detection and 97.54% for leaf pose estimation, and the average success rate of leaf grasping is 90.55%. This study presents an innovative method for expanding the applicability in vivo contact-based hyperspectral imaging for extensive agricultural applications.</div></div>","PeriodicalId":50627,"journal":{"name":"Computers and Electronics in Agriculture","volume":"236 ","pages":"Article 110483"},"PeriodicalIF":7.7,"publicationDate":"2025-05-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143941456","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":"Estimating forage mass in Brazilian pasture-based livestock production systems through satellite and climate data integration","authors":"Gustavo Bayma ,&nbsp;Sandra Furlan Nogueira ,&nbsp;Marcos Adami ,&nbsp;Edson Eyji Sano ,&nbsp;Daniel Coaguila Nuñez ,&nbsp;Patrícia Menezes Santos ,&nbsp;José Ricardo Macedo Pezzopane ,&nbsp;Célia Regina Grego ,&nbsp;Antônio Heriberto de Castro Teixeira ,&nbsp;Sergii Skakun","doi":"10.1016/j.compag.2025.110496","DOIUrl":"10.1016/j.compag.2025.110496","url":null,"abstract":"<div><div>Grasslands are vital for global food security, making reliable monitoring of forage mass (FM) essential for sustainable pasture management. The availability and quality of FM are key factors in determining the profitability of pasture-based farms. This study presents a replicable methodology for estimating FM using multi-sensor satellite data and an agrometeorological modeling framework. Conducted at the Brazilian Agricultural Research Corporation Southeast Livestock Center (Embrapa Pecuária Sudeste) in São Carlos, Brazil, the research integrates NASA’s Harmonized Landsat and Sentinel-2 (HLS) imagery with climate data processed through the Simple Algorithm for Evapotranspiration Retrieving (SAFER) and Monteith’s Light Use Efficiency (LUE) models. The SAFER model explained over 67 % of FM variability in three pasture-based livestock systems. A key factor in achieving accurate FM estimates was the differentiation between field green matter (GM) and total dry matter, as GM represents the most nutritious and consumable forage component. The model performed best in extensive systems, where minimal management intervention resulted in stable forage conditions. In integrated crop-livestock systems, the accuracy remained high, though fertilization and crop residue decomposition influenced FM estimates. In intensive systems, model performance was slightly lower due to higher management variability. This study contributes to the development of automated, scalable FM assessment methods, enabling systematic pasture monitoring and data-driven grazing management. The SAFER model allowed simultaneous processing of satellite imagery and climate data, increasing the accuracy of FM estimations. Future research should explore the use of higher-resolution imagery (e.g., CBERS-4A, PlanetScope) to better capture within-field variability and consider increasing the frequency of field sampling frequency (from 32 days to 15 or even 7 days) to further improve FM estimation accuracy, particularly in intensive systems.</div></div>","PeriodicalId":50627,"journal":{"name":"Computers and Electronics in Agriculture","volume":"237 ","pages":"Article 110496"},"PeriodicalIF":7.7,"publicationDate":"2025-05-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143936321","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":"Global path planning for navigating orchard vehicle based on fruit tree positioning and planting rows detection from UAV imagery","authors":"Yang Xu ,&nbsp;Xinyu Xue ,&nbsp;Zhu Sun ,&nbsp;Wei Gu ,&nbsp;Longfei Cui ,&nbsp;Yongkui Jin ,&nbsp;Yubin Lan","doi":"10.1016/j.compag.2025.110446","DOIUrl":"10.1016/j.compag.2025.110446","url":null,"abstract":"<div><div>This paper introduces a methodology for vehicles navigation in orchard management, based on fruit trees geolocating, planting rows detection and an improved Dijkstra algorithm. A small object detection deep learning method is proposed to enhance the detection performance of fruit trees from UAV-acquired imagery map tiles, by integrating the Large Selection Kernel and Gather-Distribute feature fusion (LSK-GD) modules. A two-step plantation-row detection algorithm is established, to merge detection results from serial map tiles to global locations and cluster trees with geo-location into rows in a global view, considering the planting metrics including the spacing in-row between adjacent trees and row spacing. Based on the calculated results of planting rows, a path planning algorithm is proposed to navigate both ground and aerial orchard vehicles and perform essential management tasks in orchards. The test results show that the detection performance of LSK-GD CNN models surpasses that of other classic models. Based on the proposed methodology, the estimated tree numbers closely match the actual tree numbers (939 → 940, and 1951 → 1650). Furthermore, the estimated row numbers both remain the same as the counted ones, with a maximum angle deviation of less than 2 degrees and an average spacing deviation of less than 0.10 m. The calculated Root Mean Square Error of the automated UGV and UAV farming planning paths is less than 0.5 m. Both the calculation time and path length using the proposed method remains shorter than those using other planning methods. The overall computational times of the data mining are 27.7 and 71.02 s for two selected fields with areas of 1.34 and 2.82 acres, respectively.</div></div>","PeriodicalId":50627,"journal":{"name":"Computers and Electronics in Agriculture","volume":"236 ","pages":"Article 110446"},"PeriodicalIF":7.7,"publicationDate":"2025-05-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143941570","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":"Precision monitoring of rice nitrogen fertilizer levels based on machine learning and UAV multispectral imagery","authors":"Ming-Der Yang ,&nbsp;Yu-Chun Hsu ,&nbsp;Yi-Hsuan Chen ,&nbsp;Chin-Ying Yang ,&nbsp;Kai-Yun Li","doi":"10.1016/j.compag.2025.110523","DOIUrl":"10.1016/j.compag.2025.110523","url":null,"abstract":"<div><div>Rice is the primary food crop globally, and effective nitrogen fertilizer management is essential for optimizing yield while minimizing environmental impact. This study integrated unmanned aerial vehicle (UAV) imagery with multispectral imaging and machine learning (ML) methods to classify nitrogen levels (N levels) in rice fields. Experimental fields with various N levels (underfertilized, optimal fertilization, and overfertilized) were imaged in 2020 and 2021 by using UAVs. The captured images underwent geometric and spectral corrections, and rice pixel segmentation was performed using a decision tree classifier, which achieved a recall of 95.3 % and an overall accuracy of 88.8 %. N level classification was performed by extracting 16 spectral and structural features from the images, including color space transformations, vegetation indices, and canopy coverage. These features were input to support vector machine (SVM) and <em>k</em> nearest neighbors (KNN) models, and feature selection methods were applied to improve performance. The SVM model outperformed the KNN model, particularly in Period II, achieving an overall accuracy of 90.0 % when the chi-square feature selection method was applied. The Red Edge Ratio Vegetation Index and canopy coverage were the most informative features for classification. The integration of UAV-based multispectral imagery and ML in this study enhanced nitrogen classification accuracy and scalability. The method provides a data-driven approach for precision agriculture and sustainable fertilization management.</div></div>","PeriodicalId":50627,"journal":{"name":"Computers and Electronics in Agriculture","volume":"237 ","pages":"Article 110523"},"PeriodicalIF":7.7,"publicationDate":"2025-05-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143936319","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":"Online detection of broken and impurity rates in half-feed peanut combine harvesters based on improved YOLOv8-Seg","authors":"Man Gu ,&nbsp;Haiyang Shen ,&nbsp;Jie Ling ,&nbsp;Zhaoyang Yu ,&nbsp;Weiwen Luo ,&nbsp;Feng Wu ,&nbsp;Fengwei Gu ,&nbsp;Zhichao Hu","doi":"10.1016/j.compag.2025.110494","DOIUrl":"10.1016/j.compag.2025.110494","url":null,"abstract":"<div><div>The broken and impurity rates are key indicators for evaluating the operational quality of peanut combine harvesters. In half-feed combine harvesters, the operating components are highly integrated, and various working parameters influence the final harvest quality. Real-time monitoring of the broken and impurity rates enables timely adjustment of these parameters to achieve optimal harvesting performance and improved efficiency. To address the current lack of suitable online detection devices for half-feed peanut combine harvesters, a lightweight device was developed for real-time detection of broken and impurity rates. By optimizing the sampling mechanism, continuous collection of harvested material is realized. An improved YOLOv8n-Seg model, incorporating multi-scale feature fusion, task decomposition, deformable convolution, and attention mechanisms, was proposed to enhance detection capabilities based on the composition of the materials. The network adaptively focuses on key regions, improving segmentation accuracy and adaptability under complex conditions. Lightweight modules make the model easy to deploy and reduce computational demands. Field test results show that at a machine speed of 0.4 m/s, the model achieves better detection performance, with errors of 3.97 % and 3.72 % for broken and impurity rates, respectively. At 0.8 m/s, the average relative errors increase to 7.69 % and 7.59 %, respectively. Deploying the monitoring system on a display terminal enables detection updates every 2 s on average, supporting timely parameter adjustments by the operator or an automated control system, thereby enhancing the quality and efficiency of peanut harvesting and reducing losses.</div></div>","PeriodicalId":50627,"journal":{"name":"Computers and Electronics in Agriculture","volume":"237 ","pages":"Article 110494"},"PeriodicalIF":7.7,"publicationDate":"2025-05-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143937364","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}