[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"item-707":3,"related-707":62},{"id":4,"title":5,"url":6,"summary":7,"summary_zh":8,"content":9,"source_name":10,"source_url":8,"published_at":11,"category":12,"cover_url":8,"hotness":13,"is_selected":14,"score":15,"score_detail":16,"sources":24,"tags":26,"primary_topic":31,"secondary_topics":37,"classification":38,"search_phrases":56,"slug":59,"view_count":60,"doi":8,"paper":8,"created_at":61},707,"基于多尺度融合与动态特征交互的仓储环境马铃薯缺陷检测(MDS-DETR)","https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F16\u002F1738","海南大学联合山东理工大学针对仓储马铃薯堆叠遮挡、缺陷形态差异大问题,构建黑斑、发芽、干腐等多类缺陷数据集,设计 MCF 上下文特征整合模块与 AIFI-DyMona 实例特征交互结构并引入 Shape-IoU 回归;MDS-DETR 以 14.2M 参数、42.9G FLOPs 取得 mAP@0.5 96.3%,集成于智能分拣机器人后真实仓储分拣准确率稳定超过 95%。",null,"Logical Operator Operator\n\nSearch Text\n\nSearch Type\n\n_add\\_circle\\_outline_\n\n_remove\\_circle\\_outline_\n\n[![Image 1: agriculture-logo](https:\u002F\u002Fpub.mdpi-res.com\u002Fimg\u002Fjournals\u002Fagriculture-logo.png?d9aa6395761dfdff)](https:\u002F\u002Fwww.mdpi.com\u002Fjournal\u002Fagriculture)\n\n## Article Menu\n\nFont Type:\n\n_Arial_ _Georgia_ _Verdana_\n\nFont Size:\n\nAa Aa Aa\n\nLine Spacing:\n\n__ __ __\n\nColumn Width:\n\n__ __ __\n\nBackground:\n\nOpen Access Article\n\nby \nDanyang Lv\n\n 1,2, \nAng Zhao\n\n 2,3, \nShuo Han\n\n 2,3, \nRanbing Yang\n\n 2,3,*[](mailto:yangranbing@hainanu.edu.cn), \nGuohai Zhang\n\n 4 and \nXiaohui Yang\n\n 4\n\n1\n\nSchool of Information and Communication Engineering, Hainan University, Haikou 570228, China\n\n2\n\nKey Laboratory of Tropical Intelligent Agricultural Equipment, Hainan University, Ministry of Agriculture and Rural Affairs, Danzhou 571737, China\n\n3\n\nSchool of Mechanical and Electrical Engineering, Hainan University, Haikou 570228, China\n\n4\n\nSchool of Agricultural Engineering and Food Science, Shandong University of Technology, Zibo 255049, China\n\n*\n\nAuthor to whom correspondence should be addressed.\n\nSubmission received: 15 July 2026 \u002F Revised: 12 August 2026 \u002F Accepted: 12 August 2026 \u002F Published: 13 August 2026\n\n## Abstract\n\nAiming at the challenges of potato storage scenarios, severe target stacking and occlusion, and large variations in defect characteristics in potato storage environments, a potato defect detection method named MDS-DETR was proposed, and its deployment on a robotic sorting platform was validated. First, potato images under different illumination conditions and stacking states were collected in potato storage warehouses to construct a potato defect dataset containing defects such as black spot, sprouting, dry rot, decay, green skin, and cracking. Subsequently, according to the characteristics of object detection tasks in storage scenarios, an MCF module was designed to strengthen the extraction and integration of contextual features across different spatial scales. An AIFI-DyMona structure, namely Anchor-free Instance Feature Interaction with Dynamic Mona, was constructed to improve the stability of feature representation under different illumination conditions. In addition, the Shape-IoU regression strategy was incorporated to improve the network sensitivity to irregular defect contours and geometric characteristics. Validation experiments demonstrated that the proposed MDS-DETR framework achieved 96.3% mAP@0.5, with only 14.2 M parameters and 42.9 G FLOPs. Compared with several representative object detection algorithms, the proposed method exhibited superior recognition capability and stronger robustness under complicated storage conditions, while also showing better suppression of missed and incorrect detections. To further evaluate its practical applicability, the trained network was integrated into an intelligent potato sorting robot and tested in real warehouse scenarios. Experimental observations indicated that the robotic system consistently maintained a sorting accuracy exceeding 95%, demonstrating the effectiveness and practical deployment potential of the proposed approach for potato storage applications. This study can provide a reference for intelligent detection and automated sorting in potato storage processes.\n\n## 1. Introduction\n\nPotato is recognized as an essential crop in global agriculture due to its high productivity, wide environmental adaptability, and abundant nutritional content [[1](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F16\u002F1738#B1-agriculture-16-01738)]. Driven by advances in large-scale cultivation methods and centralized storage systems, potatoes are increasingly preserved in cold storage facilities or ventilated warehouses after harvest for medium to long-term storage, thereby supporting off-season marketing and maintaining a stable market supply. However, during the storage process, fluctuations in temperature, humidity, and ventilation conditions can easily lead to quality problems such as decay and sprouting [[2](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F16\u002F1738#B2-agriculture-16-01738)]. The occurrence of surface damage and pathological symptoms can significantly decrease the marketability of potatoes while increasing the risk of cross-infection during storage, which further threatens the overall quality and safety of stored products [[3](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F16\u002F1738#B3-agriculture-16-01738)]. Therefore, efficient and accurate quality detection of potatoes during the storage stage is of great importance for reducing storage losses and improving economic benefits.\n\nTraditional crop quality inspection mainly relies on manual judgment, where operators identify defects such as decay and sprouting through visual observation [[4](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F16\u002F1738#B4-agriculture-16-01738)]. Although this approach offers a certain degree of flexibility, its reliance on manual inspection leads to considerable subjectivity and labor consumption. To improve detection efficiency, some studies have adopted traditional image processing techniques to identify crop appearance characteristics through the analysis of visual attributes, including chromatic, textural, and contour information. To overcome the limitations of conventional manual grading in citrus inspection, Gong et al. [[5](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F16\u002F1738#B5-agriculture-16-01738)] developed an automated defect recognition approach using machine vision technology, which significantly enhanced sorting speed and detection reliability. The method achieved defect extraction through Hue, Saturation, and Value (HSV) color space transformation, edge detection, and morphological processing, obtaining an overall recognition accuracy of 92%. Li et al. [[6](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F16\u002F1738#B6-agriculture-16-01738)] presented an automated detection approach for pineapple, which combined color and depth information with multiple image processing techniques to segment damaged regions and quantitatively calculate the damaged area through binarization processing. Nevertheless, these approaches are highly susceptible to lighting conditions and environmental noise, which limits their stability in complicated storage scenes and reduces their ability to accurately detect irregular defects, including rot and sprouting in potatoes.\n\nDriven by the rapid advancement of smart agriculture and digital farming technologies, neural network-based visual recognition methods have been applied in agricultural product inspection and quality evaluation [[7](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F16\u002F1738#B7-agriculture-16-01738)]. Models represented by convolutional neural networks and Transformers can automatically learn multi-level feature representations and have demonstrated significant advantages in quality assessment applications [[8](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F16\u002F1738#B8-agriculture-16-01738),[9](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F16\u002F1738#B9-agriculture-16-01738)]. In recent years, intelligent visual recognition techniques have been increasingly introduced into potato inspection tasks. For instance, Wang et al. [[10](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F16\u002F1738#B10-agriculture-16-01738)] established a dedicated potato image dataset and conducted comparative experiments on three convolutional neural network architectures, demonstrating the feasibility of deep learning for identifying potato defects. To improve the accuracy of potato appearance inspection, Liao et al. [[11](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F16\u002F1738#B11-agriculture-16-01738)] designed an external quality detection model named HCRP-YOLO, which achieved 95.8% mean average precision in experimental evaluations. However, most existing studies focus on ideal illumination conditions and controllable indoor scenarios, while research on applications in storage environments remains relatively limited.\n\nIn real storage scenarios, potatoes are usually stacked or scattered, resulting in mutual occlusion among individuals. In addition, soil may adhere to the potato surface and color changes may occur, leading to blurred object boundaries and making stable segmentation difficult for traditional methods. Meanwhile, illumination conditions in storage warehouses are often uneven or weak, which can easily produce shadows or locally dark regions, thereby affecting image quality. Furthermore, as storage time increases, the appearance characteristics of potatoes dynamically change, such as gradual sprout growth and expansion of decayed areas. Therefore, achieving high-precision identification of decayed and sprouted potatoes under complex storage environments has become a critical issue in potato storage management.\n\nThis study focuses on potato defect detection tasks in storage scenarios. The primary work and innovations of this study can be summarized as follows:\n\n(1) A potato image dataset close to real storage environments was constructed, covering different illumination conditions, stacking states, and defect types.\n\n(2) The Real-Time Detection Transformer (RT-DETR) framework was further refined to strengthen its ability to capture intricate feature information, leading to more reliable recognition performance under weak illumination and severe occlusion conditions.\n\n(3) The developed detection algorithm was integrated into an intelligent potato sorting robot, and practical experiments in warehouse environments were carried out to assess its real-world identification and sorting capability.\n\n## 2. Materials and Methods\n\n### 2.1. Research Framework\n\n[Figure 1](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F16\u002F1738#fig_body_display_agriculture-16-01738-f001) illustrates the workflow of this study, which mainly includes potato image collection and dataset establishment, network optimization and training, comparative evaluation of different detection models, and practical verification through robotic sorting experiments.\n\n**Figure 1.** Overall framework.\n\n### 2.2. Potato Image Collection and Dataset Preparation\n\nThe potato images were collected in March 2026 from a controlled atmosphere potato storage warehouse located in Guyuan County, Zhangjiakou City, Hebei Province, China, as shown in [Figure 2](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F16\u002F1738#fig_body_display_agriculture-16-01738-f002). The stored potato varieties mainly included Xisen 3 and Xisen 6. Image acquisition was performed using a BNC100 industrial camera (Beijing Xinhong Teaching Equipment Co., Ltd., Beijing, China). The BNC100 industrial camera is equipped with a 1\u002F2.7-inch CMOS sensor, providing an image resolution of 1600 × 1200 pixels.\n\n**Figure 2.** Data acquisition site.\n\nTo improve the adaptability of the dataset to real storage environments, complex warehouse scene characteristics were fully considered during the image acquisition process. The collected samples included both stacked and scattered potatoes. In addition, the dataset included both normal and low-light conditions, allowing the model to achieve more stable performance across diverse lighting environments. The collected defect types mainly included black spot, sprouting, dry rot, rot, greening, and cracking. Additionally, some images were obtained from the Internet.\n\nAfter image acquisition, samples with severe blur, heavy occlusion, or incomplete information were manually discarded during the data screening process, and the remaining images were annotated in LabelImg for subsequent model training. Since the primary objective of this study was defect detection for stored potatoes, multiple defect types, including black spot, sprouting, dry rot, rot, greening, and cracking, were consolidated into a unified category labeled “Defect” to alleviate the impact of category variability and enhance the model’s capability to identify defective potatoes under complex storage conditions. Normal potatoes were not assigned a separate category and were not annotated.\n\nFinally, a potato defect dataset containing 3540 images was constructed. The dataset construction process is shown in [Figure 3](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F16\u002F1738#fig_body_display_agriculture-16-01738-f003).\n\n**Figure 3.** Dataset construction process.\n\n### 2.3. MDS-DETR Detection Method\n\nProposed by the Baidu research team [[12](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F16\u002F1738#B12-agriculture-16-01738)], RT-DETR is a Transformer-based real-time detection framework that achieves faster inference and higher detection efficiency than conventional DETR architectures while preserving excellent detection accuracy. Although RT-DETR has clear advantages in end-to-end detection frameworks and inference efficiency, it still has certain limitations in storage environments. Specifically, the original backbone network has insufficient representation capability for fine-grained textures and small defects. In addition, the feature interaction process is sensitive to illumination variations, and the feature perception mechanism cannot adequately describe the geometric shape information of targets. These factors may lead to decreased recognition accuracy in potato defect detection tasks.\n\nThe proposed method optimizes the network architecture through targeted optimization of feature representation, information interaction, and feature perception to enhance overall detection performance. Based on these improvements, an enhanced detection model named MDS-DETR is constructed, as shown in [Figure 4](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F16\u002F1738#fig_body_display_agriculture-16-01738-f004).\n\n**Figure 4.** Overall structure of MDS-DETR.\n\n#### 2.3.1. MCF\n\nIn the baseline RT-DETR model, the BasicBlock is mainly responsible for feature extraction. However, this structure primarily depends on local convolution operations, and its effective receptive field is relatively limited. As a result, the network has weakened its discrimination capability when dealing with potatoes with similar appearances or complex backgrounds.\n\nAs shown in [Figure 5](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F16\u002F1738#fig_body_display_agriculture-16-01738-f005), this study introduces a Multi-scale Context Fusion module named MCF into the backbone network as a substitute for the conventional BasicBlock structure. The purpose of this module is to strengthen the extraction and integration of contextual information at different scales. On the basis of the residual learning framework, MCF incorporates the Context Guided Block (GCBlock) [[13](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F16\u002F1738#B13-agriculture-16-01738)], so that local detail information and broader semantic context can be jointly modeled. GCBlock is composed of four functional components, including a local feature extractor f loc(*), a surrounding context extractor f sur(*), a joint feature extractor f joi(*), and a global context extractor f glo(*). The module mainly includes a feature learning stage and a feature refinement stage.\n\n**Figure 5.** Structure of MCF: (**a**) MCF; (**b**) GCBlock.\n\nIn the feature learning stage, f loc(*) focuses on capturing fine-grained spatial details from neighboring regions, whereas f sur(*) is responsible for modeling broader contextual dependencies and background semantic information around the target area. Specifically, f loc(*) employs a convolution to capture local spatial features from neighboring regions, while f sur(*) uses a channel-wise dilated convolution to establish correlations between potato targets and their surrounding background. After that, the two branches are integrated by f joi(*). The calculation process is expressed as follows.\n\n$$\nf_{j o i n t} = f_{j o i} \\left(f_{l o c} , f_{s u r}\\right) = PReLU \\left(BN \\left[f_{l o c} ; f_{s u r}\\right]\\right)\n$$\n\n(1)\n\nIn the feature recalibration stage, f glo(*) is introduced by drawing inspiration from SENet. Adaptive feature weighting is achieved through a channel attention mechanism. First, the channel descriptor is generated:\n\n$$\nz = G A P \\left(f_{j o i n t}\\right) \\in R^{B \\times C}\n$$\n\n(2)\n\nThen, the channel attention weights are calculated:\n\n$$\ns = \\sigma \\left(W_{2} \\delta \\left(W_{1} z\\right)\\right)\n$$\n\n(3)\n\nwhere σ(·) denotes the Sigmoid function, δ(·) denotes the ReLU activation function, W 1∈R C\u002Fr×C and W 2∈R C×C\u002Fr represent the weight matrices, and r denotes the channel reduction ratio.\n\nThe generated attention coefficients are subsequently applied to the original feature map through channel-wise weighting, allowing important feature responses to be selectively enhanced:\n\n$$\nf_{refined} = f_{joint} \\cdot s\n$$\n\n(4)\n\nThrough this mechanism, feature channels containing critical information were adaptively emphasized, whereas irrelevant responses were weakened, enabling more effective feature optimization under global contextual guidance.\n\nBy incorporating the above design, the proposed MCF module strengthens the extraction and integration of contextual features at different scales. Compared with the traditional BasicBlock, MCF simultaneously considers local details, contextual dependencies, and global semantic representations, overcoming the limitation of insufficient contextual modeling in the original backbone network.\n\n#### 2.3.2. AIFI-DyMona\n\nWithin the RT-DETR architecture, the normalization operation in the original AIFI module was fixed as Layer Normalization (LayerNorm), which cannot adaptively adjust according to dynamic changes in input features. As a result, the stability of the model under complex illumination conditions is insufficient.\n\nTo overcome the above limitation, a Dynamic Mona structure, referred to as DyMona, is introduced in this study and integrated into the Anchor-free Instance Feature Interaction module as a substitute for the original LayerNorm operation. The proposed module integrates the multi-scale feature extraction concept of Mona [[14](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F16\u002F1738#B14-agriculture-16-01738)] and introduces the DynamicTanh (DyT) [[15](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F16\u002F1738#B15-agriculture-16-01738)] structure to achieve dynamic normalization and extreme value suppression. Based on this improvement, the enhanced AIFI module is named AIFI-DyMona.\n\nDyMona was designed to enhance the fine-grained modeling capability of local features. A normalization and fusion structure was introduced at the top of DyMona. As shown in [Figure 6](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F16\u002F1738#fig_body_display_agriculture-16-01738-f006), S 1 and S 2 were employed to adjust the input proportions. In the primary feature extraction branch, depth-wise convolution operations with different kernel sizes were employed to capture spatial information under multiple receptive field ranges. The obtained multi-scale features were subsequently integrated through average fusion to achieve effective interaction and aggregation of cross-scale information. To strengthen channel interactions, a skip connection mechanism was designed in the multi-filter convolution stage. In the first stage, spatial information was strengthened by applying adaptive depth-wise convolution operations, whereas the subsequent phase utilizes dynamically weighted point-wise convolutions to reinforce the network's ability to distinguish informative channel features.\n\n**Figure 6.** Structure of AIFI-DyMona.\n\nThe two convolution operations can be expressed as follows:\n\n$$\nf_{d w} = x + a v g \\left(\\sum_{i = 1}^{3} \\omega_{d w}^{i} \\hat{\\bigotimes} x\\right)\n$$\n\n(5)\n\n$$\nf_{p w} = x + \\omega_{p w} \\bar{\\bigotimes} x\n$$\n\n(6)\n\nwhere x represents the input feature map, ω dw i denotes the convolution weight of the i-th depth-wise convolution branch, ωdw denotes the learnable weight of the point-wise convolution, and\n\n$\\hat{\\bigotimes}$\nand\n\n$\\bigotimes$\nrepresent depth-wise convolution and point-wise convolution, respectively. Afterward, nonlinear feature representation was strengthened using the GeLU activation function, followed by an up-projection process to recover the original channel dimensionality. The overall computational process of DyMona can be expressed as follo","Agriculture 2026, 16(16), 1738 (MDPI) 2026-08-13","2026-08-13T00:00:00Z","论文",10,false,74,{"impact":17,"substance":18,"depth":19,"authority":20,"freshness":21,"relevant":22,"comment":23},15,22,18,14,5,1,"针对马铃薯仓储缺陷检测提出新方法，具有实际应用价值，但影响范围限于细分领域。",[25],{"name":10,"url":6},[27,28,29,30],"农业机器人","深度学习","缺陷检测","智慧物流",{"code":32,"slug":33,"name":34,"path":35,"parent_code":36},"A5","tubers-grains","薯类与杂粮","\u002Ftopic\u002Ffield-crops\u002Ftubers-grains","A",[],{"status":39,"locked":14,"version":40,"method":41,"evidence":42},"classified","2026-10-22.2","rules",{"candidates":43,"agricultural_evidence":46},[44,51],{"code":32,"score":45,"supported":46,"title":47,"abstract":49,"tags":50},11,true,[48],"马铃薯",[48],[],{"code":52,"score":22,"supported":14,"title":53,"abstract":54,"tags":55},"I2",[],[],[28],[57,58],"农业机器人 智慧物流 深度学习 缺陷检测","农业机器人 智慧物流","农业机器人智慧物流深度学习缺陷检测-707",4,"2026-08-18T23:59:19.116505Z",{"total":63,"page":22,"page_size":63,"items":64},6,[65,141,207,274,314,362],{"id":66,"title":67,"url":68,"summary":69,"summary_zh":70,"content":8,"source_name":71,"source_url":68,"published_at":72,"category":12,"cover_url":8,"hotness":13,"is_selected":14,"score":73,"score_detail":74,"sources":78,"tags":80,"primary_topic":84,"secondary_topics":85,"classification":86,"search_phrases":109,"slug":112,"view_count":95,"doi":113,"paper":114,"created_at":140},1709,"EOL-DQN: An intelligent cutting decision of potato seed tubers based on improved deep reinforcement learning","https:\u002F\u002Fdoi.org\u002F10.1016\u002Fj.compag.2026.112372","EOL-DQN: An intelligent cutting decision of potato seed tubers based on improved deep reinforcement learning。Computers and Electronics in Agriculture","EOL-DQN：基于改进深度强化学习的马铃薯种薯智能切块决策。Computers and Electronics in Agriculture","Computers and Electronics in Agriculture","2026-09-05T00:00:00Z",75,{"impact":17,"substance":75,"depth":19,"authority":20,"freshness":76,"relevant":22,"comment":77},20,8,"提出基于改进深度强化学习的马铃薯种薯切块决策方法，属于智慧农业细分领域的技术创新，对种业智能化有参考价值。",[79],{"name":71,"url":68},[81,82,48,83,28],"智慧农业","农业人工智能","种业",{"code":32,"slug":33,"name":34,"path":35,"parent_code":36},[],{"status":39,"locked":14,"version":40,"method":41,"evidence":87},{"candidates":88,"agricultural_evidence":46},[89,94,99,104],{"code":32,"score":17,"supported":46,"title":90,"abstract":92,"tags":93},[91],"potato",[48,91],[48],{"code":52,"score":95,"supported":14,"title":96,"abstract":97,"tags":98},2,[],[],[82,28],{"code":100,"score":22,"supported":14,"title":101,"abstract":102,"tags":103},"F1",[],[],[83],{"code":105,"score":22,"supported":14,"title":106,"abstract":107,"tags":108},"I1",[],[],[81],[110,111],"农业人工智能 智慧农业 深度学习 马铃薯","农业人工智能 智慧农业","农业人工智能智慧农业深度学习马铃薯-1709","10.1016\u002Fj.compag.2026.112372",{"doi":113,"openalex_id":115,"authors":116,"venue":71,"cited_by_count":132,"oa_url":8,"card":133,"direction":137,"ingested_from":139},"W7208757680",[117,119,121,123,125,127,129],{"name":118,"orcid":8},"Xuyang Wang",{"name":120,"orcid":8},"Yun Liu",{"name":122,"orcid":8},"Tianshuo Su",{"name":124,"orcid":8},"Yulu Sun",{"name":126,"orcid":8},"Binheng Huang",{"name":128,"orcid":8},"Yangqian Zhang",{"name":130,"orcid":131},"Wanzhi Zhang","https:\u002F\u002Forcid.org\u002F0000-0002-6196-8037",0,{"tldr":134,"method":135,"finding":136,"direction":137,"opportunity":138},"提出EOL-DQN改进深度强化学习模型，用于马铃薯种薯切块智能决策。","改进深度强化学习（DQN），结合视觉识别与决策模型。","EOL-DQN能有效提高切块决策的准确性和效率。","农业人工智能与决策模型","可探索将强化学习应用于其他农产品加工或农业机器人操作，优化决策策略。","openalex","2026-09-05T23:30:01.927676Z",{"id":142,"title":143,"url":144,"summary":145,"summary_zh":8,"content":8,"source_name":71,"source_url":144,"published_at":146,"category":12,"cover_url":8,"hotness":13,"is_selected":14,"score":147,"score_detail":148,"sources":151,"tags":153,"primary_topic":156,"secondary_topics":157,"classification":158,"search_phrases":176,"slug":179,"view_count":60,"doi":180,"paper":181,"created_at":206},1205,"Lightweight real-time detection of sweet potato virus disease by a hybrid CNN–Transformer network","https:\u002F\u002Fdoi.org\u002F10.1016\u002Fj.compag.2026.112319","Timely and precise detection of sweet potato virus disease (SPVD) is essential for ensuring crop health and improving yield quality in precision agriculture. However, conventional deep learning models often struggle to capture subtle lesion features under complex field conditions. This paper proposes Swin-SPVDNet, a hybrid CNN-Transformer detection framework with task-specific architectural enhancements that combines convolutional neural networks (CNNs) with a modified Swin Transformer to effectively integrate local spatial details and global contextual information. Unlike conventional hybrid models that simply stack convolutional and transformer blocks, the proposed network introduces three key structural designs: (1) A C2PSA-enhanced convolutional backbone for improved local feature sensitivity. (2) A C3STR module that tightly integrates convolutional operations with Swin Transformer blocks for joint local–global feature interaction. (3) A Residual Attention Block (RAB) that refines feature representations while stabilizing gradient propagation. A field-collected SPVD image dataset was constructed using handheld RGB imaging under real agricultural conditions. Extensive experiments and ablation studies demonstrate that Swin-SPVDNet achieves competitive performance compared with representative state-of-the-art detectors while providing a favorable balance between detection accuracy and deployment efficiency, attaining a mean Average Precision (mAP@0.5) of 86.8%, a precision of 85.1%, and a recall of 75.6%. Compared to Yolov5-v11 models, our approach demonstrates improved robustness in detecting small and irregular lesions, particularly under occlusion and lighting variability. These results confirm the practical value of Swin-SPVDNet for real-time field deployment and support its application in intelligent disease monitoring.","2026-08-25T00:00:00Z",73,{"impact":19,"substance":18,"depth":19,"authority":149,"freshness":95,"relevant":22,"comment":150},13,"提出轻量级混合CNN-Transformer模型用于甘薯病毒病实时检测，方法新颖且实验数据详实，对智能植保有参考价值。",[152],{"name":71,"url":144},[81,82,154,28,155],"作物病害","甘薯",{"code":32,"slug":33,"name":34,"path":35,"parent_code":36},[],{"status":39,"locked":14,"version":40,"method":41,"evidence":159},{"candidates":160,"agricultural_evidence":46},[161,166,171],{"code":32,"score":162,"supported":46,"title":163,"abstract":164,"tags":165},12,[91],[91],[155],{"code":52,"score":21,"supported":46,"title":167,"abstract":168,"tags":170},[],[169],"deep learning",[82,28],{"code":105,"score":60,"supported":46,"title":172,"abstract":173,"tags":175},[],[174],"precision agriculture",[81],[177,178],"农业人工智能 作物病害 智慧农业 深度学习","农业人工智能 作物病害","农业人工智能作物病害智慧农业深度学习-1205","10.1016\u002Fj.compag.2026.112319",{"doi":180,"openalex_id":182,"authors":183,"venue":71,"cited_by_count":132,"oa_url":144,"card":201,"direction":137,"ingested_from":139},"W7204154634",[184,187,190,193,195,197,199],{"name":185,"orcid":186},"Ru Han","https:\u002F\u002Forcid.org\u002F0009-0000-5128-1314",{"name":188,"orcid":189},"Lei Shu","https:\u002F\u002Forcid.org\u002F0000-0002-6700-9347",{"name":191,"orcid":192},"Grzegorz Cielniak","https:\u002F\u002Forcid.org\u002F0000-0002-6299-8465",{"name":194,"orcid":8},"Yan Zhu",{"name":196,"orcid":8},"Xing Yang",{"name":198,"orcid":8},"Fan Yang",{"name":200,"orcid":8},"Kai Huang",{"tldr":202,"method":203,"finding":204,"direction":137,"opportunity":205},"提出Swin-SPVDNet混合CNN-Transformer网络，实现甘薯病毒病轻量级实时检测。","结合C2PSA卷积骨干、C3STR模块和残差注意力块，使用手持RGB图像数据集。","mAP达86.8%，精度85.1%，召回75.6%，优于YOLOv5-v11，鲁棒性强。","可探索将混合网络应用于其他作物病害检测，或优化模型以适应更多复杂田间环境。","2026-09-01T04:03:02.220802Z",{"id":208,"title":209,"url":210,"summary":211,"summary_zh":212,"content":8,"source_name":213,"source_url":210,"published_at":214,"category":12,"cover_url":8,"hotness":13,"is_selected":14,"score":215,"score_detail":216,"sources":220,"tags":222,"primary_topic":226,"secondary_topics":227,"classification":228,"search_phrases":252,"slug":255,"view_count":132,"doi":256,"paper":257,"created_at":273},5561,"Growth Stage and Cultivar Influence the Interpretation of UAV-Derived NDVI for Potato Yield Assessment","https:\u002F\u002Fdoi.org\u002F10.1007\u002Fs11540-026-10161-5","Abstract Single-date UAV vegetation indices are widely used to assess potato crop performance and estimate yield. However, whether their biological interpretation remains consistent across cultivars and growth stages is rarely examined. This study investigated how the relationships between UAV-derived multispectral and thermal canopy indicators and potato yield changed during crop development under contrasting irrigation in two cultivars with contrasting maturity classes (Anouk and Celandine). UAV surveys were conducted in July and August, and canopy metrics were related to final yield and tuber traits. The relationship between NDVI and yield differed markedly between cultivars and growth stages. In the early-maturing cultivar Anouk, July NDVI showed a strong positive correlation with yield ( r = 0.87, p = 0.002), whereas August NDVI was not informative. In the late-maturing cultivar Celandine, the relationship reversed from a positive correlation in July ( r = 0.56) to a strong negative correlation in August ( r = −0.80, p = 0.009), indicating that greater late-season canopy greenness was associated with lower yield. Exploratory LOOCV Random Forest models achieved the highest yield prediction accuracy using July features for Anouk ( R 2 = 0.72) and August features for Celandine ( R 2 = 0.47), while multi-temporal features provided similar predictive performance. Greater temporal declines in NDVI were associated with higher yield in both cultivars, consistent with the change in canopy greenness acting as a phenological indicator of the transition from vegetative growth to tuber bulking. These results demonstrate that the interpretation of UAV-derived vegetation indices is strongly dependent on cultivar maturity and growth stage. A single UAV acquisition may therefore provide misleading assessments of crop performance when canopy phenology is not considered. Rather than serving as fixed proxies for yield, canopy vegetation indices should be interpreted within cultivar-specific and temporal contexts when used for potato monitoring and agronomic decision-making.","摘要 单期无人机植被指数被广泛用于评估马铃薯作物表现和估算产量。然而，其生物学解释在不同品种和生育期之间是否保持一致，鲜有研究加以检验。本研究探讨了在两种成熟性类别不同的品种（Anouk和Celandine）中，不同灌溉条件下，无人机获取的多光谱与热红外冠层指标与马铃薯产量之间的关系在作物发育过程中如何变化。无人机调查于7月和8月进行，并将冠层指标与最终产量及块茎性状进行关联分析。NDVI与产量之间的关系在品种和生育期之间差异显著。在早熟品种Anouk中，7月NDVI与产量呈强正相关（r = 0.87，p = 0.002），而8月NDVI则不具有指示意义。在晚熟品种Celandine中，该关系由7月的正相关（r = 0.56）转变为8月的强负相关（r = −0.80，p = 0.009），表明后期冠层绿度越高与产量越低相关。探索性留一交叉验证（LOOCV）随机森林模型在使用Anouk的7月特征（R² = 0.72）和Celandine的8月特征（R² = 0.47）时取得了最高的产量预测精度，而多时相特征提供了相似的预测表现。两个品种中NDVI随时间下降幅度越大均与产量越高相关，这与冠层绿度变化作为从营养生长向块茎膨大转变的物候指示相一致。这些结果表明，无人机获取的植被指数的解释强烈依赖于品种成熟性和生育期。因此，在未考虑冠层物候的情况下，单次无人机采集可能对作物表现作出误导性评估。冠层植被指数不应作为产量的固定替代指标，而应在品种特异性和时间背景下加以解释，方可用于马铃薯监测和农艺决策。","Potato Research","2026-10-20T00:00:00Z",79,{"impact":217,"substance":18,"depth":19,"authority":20,"freshness":218,"relevant":22,"comment":219},16,9,"该研究揭示UAV-NDVI与马铃薯产量的关系随品种熟期和生育阶段而反转，对智慧农业遥感监测具有实质方法学与决策参考价值。",[221],{"name":213,"url":210},[81,48,223,224,225],"产量预测","遥感","作物表型",{"code":32,"slug":33,"name":34,"path":35,"parent_code":36},[],{"status":39,"locked":14,"version":40,"method":41,"evidence":229},{"candidates":230,"agricultural_evidence":46},[231,235,242,248],{"code":32,"score":17,"supported":46,"title":232,"abstract":233,"tags":234},[91],[48,91],[48],{"code":236,"score":63,"supported":46,"title":237,"abstract":238,"tags":241},"H3",[],[239,240],"灌溉","irrigation",[],{"code":243,"score":63,"supported":46,"title":244,"abstract":245,"tags":247},"I3",[],[246],"multispectral",[223,225,224],{"code":105,"score":22,"supported":14,"title":249,"abstract":250,"tags":251},[],[],[81],[253,254],"UAV NDVI 马铃薯 产量","马铃薯 品种 生育期 遥感","UAVNDVI马铃薯产量-5561","10.1007\u002Fs11540-026-10161-5",{"doi":256,"openalex_id":258,"authors":259,"venue":213,"cited_by_count":132,"oa_url":265,"card":266,"direction":272,"ingested_from":139},"W7223067378",[260,263],{"name":261,"orcid":262},"Jiangsan Zhao","https:\u002F\u002Forcid.org\u002F0000-0003-3916-7388",{"name":264,"orcid":8},"Jaroslaw S. Grodek","https:\u002F\u002Flink.springer.com\u002Fcontent\u002Fpdf\u002F10.1007\u002Fs11540-026-10161-5.pdf",{"tldr":267,"method":268,"finding":269,"direction":270,"opportunity":271},"研究不同品种和生育期下无人机NDVI与马铃薯产量的关系差异。","两个熟期品种、多时相无人机多光谱与热红外冠层指标，结合随机森林建模。","NDVI与产量的关系随品种和时期反转，晚熟品种8月NDVI与产量强负相关。","农业遥感与作物表型","可构建融合品种熟期与冠层物候期的时相自适应产量估算模型，避免单次飞行误判。","智慧农业 \u002F 农业物联网","2026-10-22T11:20:21.975818Z",{"id":275,"title":276,"url":277,"summary":278,"summary_zh":8,"content":279,"source_name":280,"source_url":8,"published_at":214,"category":281,"cover_url":8,"hotness":13,"is_selected":14,"score":147,"score_detail":282,"sources":284,"tags":286,"primary_topic":291,"secondary_topics":292,"classification":293,"search_phrases":309,"slug":312,"view_count":132,"doi":8,"paper":8,"created_at":313},5526,"甘肃农科院陇薯红1号千亩方实收测产 亩产3033.59公斤商品率92%","https:\u002F\u002Fmlss.gsagr.cn\u002Finfo\u002F1056\u002F2213.htm","10月18日马铃薯研究所科研团队在会宁县大沟镇郭庄村陇薯红1号千亩示范方开展现场测产，实测亩产3033.59公斤、商品率92%。该品种为鲜食与加工兼用型，薯皮光滑呈红色、薯肉红芯黄边富含花青素，抗旱性与综合抗逆性突出，尤宜炸制薯片薯条，已获多家薯片薯条加工企业认可。","测产现场，专家组严格按照马铃薯田间测产技术规程，对自主选育的陇薯红1号进行实收测产，并综合评价其农艺性状、抗逆性与加工品质。经实测，该品种田间长势整齐，丰产性突出，折合亩产3033.59公斤，商品率达92%，抗逆表现良好，块茎外观品质优良，符合加工原料标准。项目负责人、马铃薯所长马丽荣介绍，陇薯红1号是我院马铃薯研究所最新选育的鲜食与加工兼用型品种，尤宜炸制薯片、薯条。该品种薯形规整，薯皮光滑呈红色，薯肉红芯黄边，富含花青素，抗旱性与综合抗逆性突出，炸制的薯片色泽鲜亮、片形均匀、品质稳定。该品种的育成进一步拓展了功能加工型马铃薯的育种方向与应用场景，已获得省内外多家薯片、薯条加工企业的高度评价与广泛认可。","甘肃省农业科学院马铃薯研究所 2026-10-20","报道",{"impact":19,"substance":75,"depth":20,"authority":149,"freshness":76,"relevant":22,"comment":283},"省级科研机构自主选育的功能加工型马铃薯品种千亩方实收测产，亩产与商品率数据具体，兼具鲜食加工兼用与花青素特色，具备行业参考价值。",[285],{"name":280,"url":277},[48,287,288,289,290],"种业振兴","测产验收","功能加工型","陇薯红1号",{"code":32,"slug":33,"name":34,"path":35,"parent_code":36},[],{"status":39,"locked":14,"version":40,"method":41,"evidence":294},{"candidates":295,"agricultural_evidence":46},[296,300,305],{"code":32,"score":60,"supported":46,"title":297,"abstract":298,"tags":299},[],[48],[48],{"code":301,"score":22,"supported":14,"title":302,"abstract":303,"tags":304},"A8",[],[],[288],{"code":100,"score":22,"supported":14,"title":306,"abstract":307,"tags":308},[],[],[287],[310,311],"甘肃农科院 陇薯红1号 测产","陇薯红1号 亩产3033.59公斤","甘肃农科院陇薯红1号测产-5526","2026-10-22T00:07:04.811596Z",{"id":315,"title":316,"url":317,"summary":318,"summary_zh":8,"content":319,"source_name":320,"source_url":8,"published_at":214,"category":281,"cover_url":8,"hotness":13,"is_selected":14,"score":321,"score_detail":322,"sources":324,"tags":326,"primary_topic":331,"secondary_topics":332,"classification":333,"search_phrases":357,"slug":360,"view_count":132,"doi":8,"paper":8,"created_at":361},5514,"山东农科院作物所商河举办红高粱观摩培训会 政企共建示范基地","http:\u002F\u002Fsdcrops.cn\u002Farticles\u002Fch11158\u002F202610\u002F32fdb8f1-6d3b-441c-8ef9-3d0d3332ce90.shtml","10月18日第四届「红高粱·庆丰收」山东省酿酒高粱新品种新技术现场观摩培训会在商河县金盆地高粱种植示范基地举办，省农科院作物所联合山东省白酒协会与商河县人民政府主办，50余名酿酒畜牧企业代表与种植大户参加。会上颁发「科技成果转化专员」聘书并签署示范基地共建协议。","秋染良田，红粱飘香。为加快酿酒高粱新品种新技术成果转化，深化院地产学研协同创新，10月18日，第四届“红高粱・庆丰收”山东省酿酒高粱新品种新技术现场观摩培训会在商河县金盆地高粱种植示范基地成功举办。本次活动由山东省农业科学院作物研究所联合山东省白酒协会、商河县人民政府共同主办，商河县委组织部、县农业农村局、县人社局、县科协、县广播电视台、龙桑寺镇人民政府承办，商河县金盆地家庭农场协办。省农科院党委委员、副院长刘开昌，作物所理事长兼党委书记赵海军，山东省白酒协会秘书长李浩然，商河县副县长任明光及县直相关部门负责同志出席；省内外酿酒企业、畜牧企业代表，部分家庭农场、合作社及高粱种植大户50余人参加。\n\n活动以酿酒高粱试验示范基地建设为核心，搭建科研成果落地田间的转化平台，集中展示作物所酿酒高粱育种与栽培技术最新成果。国家谷子高粱产业技术体系济南综合试验站站长、作物所杂粮团队首席专家张华文研究员在田间开展技术培训，围绕新品种选育进展、标准化高产栽培技术、产业发展现状与趋势等核心内容，与种植大户、企业代表现场互动交流，针对性解答生产一线实际问题，推动良种良法直送生产端。\n\n现场观摩培训会由龙桑寺镇党委书记刘友智主持，任明光、李浩然、刘开昌先后致辞。赵海军为金盆地家庭农场负责人史益民颁发“科技成果转化专员”聘书，双方共同签署酿酒高粱试验示范基地共建协议，建立长期稳定的科研合作纽带。随后，与会领导为开镰仪式剪彩，标志着示范基地高粱正式进入丰收收割阶段。与会人员还深入田间观摩酿酒高粱新品种示范田。张华文现场讲解品种特征特性、产量表现及配套轻简栽培技术，直观展示科研成果的应用成效与产业价值。\n\n本次活动是作物所深化院地合作、推进科技成果转化的重要实践。商河金盆地高粱示范基地作为院地共建核心载体，已成为科研人才下沉一线、新品种新技术落地推广的田间阵地，有效打通科研与生产的衔接通道。下一步，作物所将持续发挥科研人才与技术优势，深化与金盆地家庭农场的产学研协同，开展酿酒高粱新品种试验与栽培技术攻关，进一步发挥示范基地辐射带动作用，推动科研成果转化为产业效益，带动周边农户增收致富，为区域农业产业高质量发展注入持久科技动能。（撰写：王润丰 核稿：李峰）","山东省农业科学院作物研究所 2026-10-20",62,{"impact":217,"substance":20,"depth":162,"authority":162,"freshness":76,"relevant":22,"comment":323},"省级科研院所与地方政府共建酿酒高粱示范基地的观摩培训活动，有实质内容但属常规院地合作动态，公共影响有限。",[325],{"name":320,"url":317},[287,327,328,329,330],"成果转化","院地合作","示范基地","酿酒高粱",{"code":32,"slug":33,"name":34,"path":35,"parent_code":36},[],{"status":39,"locked":14,"version":40,"method":41,"evidence":334},{"candidates":335,"agricultural_evidence":46},[336,341,346,353],{"code":32,"score":162,"supported":46,"title":337,"abstract":339,"tags":340},[338],"高粱",[338],[330],{"code":342,"score":21,"supported":46,"title":343,"abstract":344,"tags":345},"G1",[],[327],[327,328],{"code":347,"score":348,"supported":46,"title":349,"abstract":350,"tags":352},"C",3,[],[351],"畜牧",[],{"code":100,"score":22,"supported":14,"title":354,"abstract":355,"tags":356},[],[],[287],[358,359],"山东省农科院 酿酒高粱 商河","金盆地 高粱示范基地","山东省农科院酿酒高粱商河-5514","2026-10-22T00:07:03.937590Z",{"id":363,"title":364,"url":365,"summary":366,"summary_zh":8,"content":367,"source_name":368,"source_url":8,"published_at":369,"category":281,"cover_url":8,"hotness":13,"is_selected":14,"score":73,"score_detail":370,"sources":372,"tags":374,"primary_topic":378,"secondary_topics":379,"classification":380,"search_phrases":412,"slug":415,"view_count":22,"doi":8,"paper":8,"created_at":416},5438,"河北农科院谷子所冀研谷3号在藁城实打测产 亩产 431.5 公斤 增产 13.2%","http:\u002F\u002Fhebnky.com\u002Fjishuzhidaoxinxi.aspx?jszdid=2226","河北省农林科学院谷子所、农机所联合石家庄市农技推广中心，2026 年在石家庄藁城留章建立 500 亩试验基地，为「藁城宫米」区域公用品牌储备专用品种资源。9 月 29 日专家组对冀研谷3号百亩单产提升示范田实打实收测产，亩产 431.5 公斤，较对照田增产 13.2%，亩节本增效 269.2 元。冀研谷3号是面向冀中南夏谷区选育的高产广适夏谷中熟品种，抗烯禾啶除草剂，适配麦茬夏播两熟制茬口。","藁城是谷子传统种植区和天然富硒小米优势产区，“藁城宫米”于2020年获中国国家地理标志证明商标，是河北省名优农产品区域公用品牌。区域公用品牌的持续壮大离不开优质专用品种的稳定支撑。藁城区此前已联合谷子所筛选培育出“宫米一号”专用品种。随着产业规模扩大和市场需求多元化，藁城宫米品种谱系需持续扩充。围绕这一目标，2026年，谷子所、农机所联合石家庄市农业技术推广中心，在藁城留章建立500亩试验基地，涵盖品种展示、宽窄行穴播、条播精量播种、中耕除草、叶面肥“一喷多促”及单产提升技术。\n\n藁城地处冀中南夏谷区，小麦收获后麦茬夏播是主要种植模式，对品种提出特殊要求：生育期要适配小麦—谷子两熟制时间窗口，抗除草剂能力要解决麦茬地杂草压力大的问题，品质要满足宫米加工对籽粒外观和食味的要求。冀研谷3号正是面向这一需求选育的夏谷中熟品种，高产广适，抗烯禾啶除草剂。以冀研谷3号为载体开展品种筛选与配套技术验证，旨在为宫米品牌储备“优质+专用+适播”的后备品种资源。基地采用搭载北斗导航系统的播种机，在谷子生长中后期，实施“一喷多促”技术，通过无人机一次喷施，同步实现防病治虫、促灌浆、促籽粒增重等多重效果。冀研谷3号百亩单产提升示范田自6月21日播种以来，从种到收实施精细管理，条播精量播种配合飞防“一喷多促”，形成了“良种+良法+良机”的技术组装模式。9月29日，专家组对冀研谷3号进行实打实收测产。示范田亩产431.5公斤，较对照田增产13.2%，亩节本增效269.2元。\n\n藁城留章示范田的院地联合实践，为宫米产业集群建设提供了可复制的工作路径。品种筛选的持续深化、配套技术的集成落地、科研院所与推广机构协同机制的有效运转，正在为这一目标提供从品种到技术、从田间到品牌的系统性支撑。","河北省农林科学院 2026-10-19","2026-10-19T00:00:00Z",{"impact":19,"substance":75,"depth":217,"authority":149,"freshness":76,"relevant":22,"comment":371},"省级科研机构发布的夏谷新品种实打测产数据具体，良种良法良机配套路径清晰，对区域公用品牌建设有示范价值，值得进入每日精选。",[373],{"name":368,"url":365},[287,375,328,376,377],"单产提升","谷子","藁城宫米",{"code":32,"slug":33,"name":34,"path":35,"parent_code":36},[],{"status":39,"locked":14,"version":40,"method":41,"evidence":381},{"candidates":382,"agricultural_evidence":46},[383,387,391,396,402,408],{"code":32,"score":162,"supported":46,"title":384,"abstract":385,"tags":386},[376],[376],[376],{"code":301,"score":60,"supported":46,"title":388,"abstract":389,"tags":390},[],[375],[375],{"code":342,"score":60,"supported":46,"title":392,"abstract":393,"tags":395},[],[394],"农技推广",[328],{"code":397,"score":348,"supported":46,"title":398,"abstract":399,"tags":401},"F4",[],[400],"除草",[],{"code":403,"score":348,"supported":46,"title":404,"abstract":405,"tags":407},"F5",[],[406],"农机",[],{"code":100,"score":22,"supported":14,"title":409,"abstract":410,"tags":411},[],[],[287],[413,414],"冀研谷3号 藁城 测产","河北农科院 谷子 宫米","冀研谷3号藁城测产-5438","2026-10-21T00:10:01.499899Z"]