[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"item-842":3,"related-842":57},{"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":23,"tags":25,"primary_topic":30,"secondary_topics":36,"classification":37,"search_phrases":51,"slug":54,"view_count":55,"doi":8,"paper":8,"created_at":56},842,"多模型融合框架用于复杂果园环境下稳健芒果检测(原标题:A Multi-Model Fusion Framework for Robust Mango Detection in Complex Orchard Environments)","https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F16\u002F1770","华南农业大学工程学院Lu等提出多模型融合框架用于复杂果园环境下稳健芒果检测:以YOLOv8n、YOLOv8s和YOLOv8m为基检测器,应用多尺度测试时增强(TTA)获取不同增强视图的预测;基于交并比(IoU)将每个基检测器不同TTA视图对应同一目标的预测进行匹配;采用加权框融合(WBF)确定融合后的边界框坐标;对多个基检测器共同检测的候选目标使用Noisy-OR组合各模型置信度得分;应用高斯Soft-NMS降低密集场景下相邻芒果的错误抑制风险。在标准数据集上Recall和mAP@0.5分别达到95.52%和98.60%。",null,"## 1. Introduction\n\nWith the global population expected to exceed 9.7 billion by 2050, agricultural production must increase significantly to meet growing food demand [[1](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F16\u002F1770#B1-agriculture-16-01770)]. In this context, precision agriculture has emerged as a key paradigm, leveraging data-driven technologies to improve productivity and resource efficiency [[2](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F16\u002F1770#B2-agriculture-16-01770)]. Fruit production, particularly for high-value crops such as mango, plays a critical role in supporting agricultural economies and farmer incomes in tropical and subtropical regions [[3](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F16\u002F1770#B3-agriculture-16-01770),[4](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F16\u002F1770#B4-agriculture-16-01770)]. Accurate fruit detection and localization are fundamental to applications such as yield estimation, growth monitoring, and robotic harvesting. Vision-based sensing systems deployed on unmanned ground vehicles (UGVs) or unmanned aerial vehicles (UAVs) provide an effective solution for large-scale orchard monitoring [[5](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F16\u002F1770#B5-agriculture-16-01770),[6](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F16\u002F1770#B6-agriculture-16-01770)]. Field-deployed vision systems on unmanned ground vehicles (UGVs) and unmanned aerial vehicles (UAVs) operate under challenging imaging and computational conditions. Uneven terrain, platform vibration, motion blur, changing viewpoints, dust, and rapidly varying illumination can reduce image quality and detection stability. In addition, onboard platforms generally have limited computing resources, memory, energy supply, and thermal capacity. UAV-based systems are further constrained by payload, battery endurance, flight altitude, and communication bandwidth. These practical factors require detection methods to balance accuracy, environmental robustness, and computational efficiency according to the intended application. Accordingly, the present study prioritizes detection accuracy and robustness in complex orchard scenes, while also quantitatively evaluating the computational overhead of the proposed framework. However, despite advances in deep learning-based object detection, achieving reliable fruit detection in real-world orchard environments remains a challenging problem [[7](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F16\u002F1770#B7-agriculture-16-01770)].\n\nUnlike structured industrial environments, natural orchards exhibit significant variability and environmental uncertainty, which introduces multiple challenges for visual perception. Severe occlusion and dense fruit overlapping are common, where fruits, leaves, and branches obscure each other, leading to incomplete visual information and increased missed detections [[8](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F16\u002F1770#B8-agriculture-16-01770),[9](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F16\u002F1770#B9-agriculture-16-01770),[10](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F16\u002F1770#B10-agriculture-16-01770)]. Traditional non-maximum suppression (NMS) methods may further merge adjacent fruits, causing counting errors in clustered scenarios [[10](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F16\u002F1770#B10-agriculture-16-01770)]. In addition, visual camouflage significantly reduces the discriminative power of color features, as many mango varieties share similar color characteristics with surrounding foliage [[11](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F16\u002F1770#B11-agriculture-16-01770),[12](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F16\u002F1770#B12-agriculture-16-01770)]. Illumination variations such as backlighting, shadows, and reflections further exacerbate this issue by altering the appearance of fruits across different conditions [[13](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F16\u002F1770#B13-agriculture-16-01770),[14](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F16\u002F1770#B14-agriculture-16-01770),[15](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F16\u002F1770#B15-agriculture-16-01770),[16](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F16\u002F1770#B16-agriculture-16-01770)]. Moreover, scale variation caused by changing camera distances results in the coexistence of large and small targets in a single image, making it difficult for detection models to maintain both sensitivity and localization accuracy [[17](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F16\u002F1770#B17-agriculture-16-01770),[18](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F16\u002F1770#B18-agriculture-16-01770),[19](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F16\u002F1770#B19-agriculture-16-01770)]. These factors collectively impose stringent requirements on the robustness and generalization capability of detection methods in agricultural scenarios.\n\nEarly fruit detection approaches relied on handcrafted features combined with traditional machine learning methods. While effective in controlled environments, these approaches are highly sensitive to illumination changes and complex backgrounds, limiting their applicability in real-world orchards [[20](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F16\u002F1770#B20-agriculture-16-01770)]. With the development of deep learning, convolutional neural networks have significantly improved detection performance. Two-stage detectors such as Faster R-CNN achieve high accuracy but are computationally expensive, making them less suitable for real-time agricultural applications. Consequently, single-stage detectors, particularly the YOLO (You Only Look Once) series, have become widely adopted due to their favorable balance between accuracy and efficiency [[8](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F16\u002F1770#B8-agriculture-16-01770)]. Recent studies have focused on improving YOLO-based models by introducing attention mechanisms, feature fusion modules, and lightweight architectures, which enhance performance under specific conditions such as occlusion or small targets [[21](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F16\u002F1770#B21-agriculture-16-01770),[22](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F16\u002F1770#B22-agriculture-16-01770),[23](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F16\u002F1770#B23-agriculture-16-01770),[24](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F16\u002F1770#B24-agriculture-16-01770)]. However, most existing approaches still rely on single-model architectures, which are inherently limited in handling the diverse and long-tail variations present in complex orchard environments.\n\nTo address these limitations, ensemble learning and inference-level enhancement strategies have been explored. Techniques such as test-time augmentation (TTA), weighted box fusion (WBF), and Soft-NMS have been shown to improve detection robustness by aggregating predictions from multiple views or models [[25](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F16\u002F1770#B25-agriculture-16-01770),[26](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F16\u002F1770#B26-agriculture-16-01770)]. Nevertheless, these techniques are often applied independently, lacking a unified framework that systematically integrates multi-model diversity and decision-level fusion for agricultural scenarios. Therefore, there is a need for a more comprehensive detection paradigm that can effectively exploit complementary information across models while maintaining robustness under complex environmental conditions.\n\nIn this study, we propose a multi-model fusion framework for robust mango detection in complex orchard environments. TTA, WBF, Noisy-OR, and Soft-NMS are established techniques. The methodological contribution of this study lies in their task-oriented integration into a unified multi-model and multi-view inference framework, together with an aggregation strategy that avoids repeatedly counting correlated TTA predictions from the same detector. The proposed approach leverages multiple detection models with different capacities and integrates their predictions through a unified inference pipeline to improve robustness and reliability. The main contributions of this work are summarized as follows: (1) a multi-model fusion strategy is developed to exploit complementary strengths across models, improving detection robustness under complex agricultural conditions; (2) a unified inference pipeline is designed by integrating multi-scale augmentation and decision-level fusion techniques, enhancing stability under occlusion, illumination variation, and dense object overlap; and (3) experiments on two complementary datasets demonstrate the effectiveness of the proposed framework compared with single-model baselines under within-dataset evaluation protocols.\n\n## 2. Materials and Methods\n\n### 2.1. Overview of the Proposed Framework\n\nAs illustrated in [Figure 1](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F16\u002F1770#fig_body_display_agriculture-16-01770-f001), the proposed mango detection framework consists of five stages: multi-scale test-time augmentation (TTA) inference, inverse transformation mapping and within-model TTA matching and fusion, cross-model matching and WBF-based coordinate fusion, model-level consensus rescoring based on Noisy-OR, and Gaussian Soft-NMS post-processing.\n\n**Figure 1.** Overall architecture and workflow of the proposed framework. The arrows indicate the sequential data flow between modules, while the different colors are used to visually distinguish the five processing stages and do not encode additional quantitative information.\n\n(1)\nMulti-scale TTA inference.\n\nThe input orchard image is transformed using multi-scale resizing and horizontal flipping to generate six TTA views. These views are independently processed by three base detectors, namely YOLOv8n, YOLOv8s, and YOLOv8m, to obtain predictions under different test-time transformations.\n\n(2)\nInverse transformation mapping and within-model TTA matching and fusion.\n\nThe predicted bounding boxes from the different TTA views are first mapped back to the coordinate system of the original image through the corresponding inverse transformations. Within each base detector, predictions of the same class are then matched according to a predefined TTA matching threshold based on IoU. Predictions considered to correspond to the same target are fused to produce a single model-specific candidate detection for subsequent cross-model fusion.\n\n(3)\nCross-model matching and WBF-based coordinate fusion.\n\nThe model-specific candidate detections produced by YOLOv8n, YOLOv8s, and YOLOv8m are matched across base detectors according to class labels and a predefined cross-model matching threshold. Candidates considered to correspond to the same target are assigned to the same cross-model candidate cluster. After removing repeated contributions from the same detector, only candidate clusters supported by at least two distinct base detectors are retained. WBF is then applied to fuse the bounding-box coordinates of the retained candidates.\n\n(4)\nModel-level consensus rescoring based on Noisy-OR.\n\nFor each retained candidate target, the model-specific confidence scores provided by the supporting base detectors are adjusted using the corresponding detector reliability weights and Score Power. The adjusted confidence scores are then combined using Noisy-OR to obtain the final consensus-ranking score. Because the three base detectors belong to the same YOLOv8 family and are trained on the same dataset, the resulting score is interpreted as a heuristic consensus-ranking score rather than as a calibrated posterior probability.\n\n(5)\nGaussian Soft-NMS post-processing.\n\nFinally, Gaussian Soft-NMS is applied to the remaining candidate boxes. Instead of directly removing highly overlapping boxes, Gaussian Soft-NMS continuously decays their confidence scores according to their overlap, thereby suppressing residual redundant detections while reducing the risk of incorrectly suppressing adjacent mangoes in densely clustered scenes.\n\n### 2.2. Selection of Base Models and Construction of Model Diversity\n\nAn effective ensemble detection method relies on prediction diversity and error complementarity among base models. To achieve a balance among computational complexity, feature representation capability, and deployment scalability, three YOLOv8 models with different parameter scales, namely YOLOv8n, YOLOv8s, and YOLOv8m, are selected as the base detectors in this study. These models share consistent input and output interfaces, which facilitates their integration within a unified post-processing framework. Meanwhile, they exhibit complementary characteristics in terms of feature learning capacity and target response behavior.\n\nYOLOv8n is a lightweight model with high inference efficiency, making it suitable for rapid detection of clear targets. However, due to its relatively limited network capacity, it may suffer from missed detections in scenarios involving severe occlusion, uneven illumination, and complex background textures.\n\nYOLOv8s is a medium-scale model that provides a favorable trade-off between inference speed and detection accuracy. It can produce relatively stable intermediate-scale feature representations and maintain good detection performance in most conventional orchard images.\n\nYOLOv8m is a larger-capacity model with deeper feature extraction layers and stronger semantic modeling capability. Therefore, it generally exhibits higher robustness when detecting small-scale fruits, partially occluded targets, and objects under complex background interference.\n\nThe combination of models with different capacities enables the detection system to simultaneously exploit the fast response of lightweight models and the complex feature representation capability of larger models. As a result, complementary detection responses can be formed under different target scales, occlusion levels, and background interference conditions, thereby improving the overall stability of the detection system.\n\n### 2.3. Test-Time Augmentation and Inverse Transformation Mapping\n\nTo alleviate the adverse effects of varying imaging distances, target scale differences, and local occlusions on detection performance, test-time augmentation (TTA) is introduced during the inference stage of the base models [[27](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F16\u002F1770#B27-agriculture-16-01770)]. Let the original input image be denoted by $I$. The proposed framework contains $M = 3$ base detectors, denoted by $f_{\\theta_{m}}$, where $m \\in \\left{\\right. 1 , \\ldots , M \\left.\\right}$ corresponds to YOLOv8n, YOLOv8s, and YOLOv8m, respectively. TTA defines a set of transformations $\\mathcal{T} = \\left{\\right. T_{1} , T_{2} , \\ldots , T_{K} \\left.\\right}$, where $K = 6$ denotes the number of test-time views. The transformations adopted in this study include scale resizing and horizontal flipping. Specifically, the scaling factors are set to 0.9, 1.0, and 1.1, corresponding to 90% down-scaling, the original scale, and 110% up-scaling, respectively. For each scaled image, both the original orientation and the horizontally flipped version are retained. Therefore, each base detector produces predictions from six test-time views.\n\nFor the $m$-th base detector under the $j$-th TTA view, the prediction set is denoted by\n\n$$\n\\mathcal{B}_{m , j} = f_{\\theta_{m}} \\left(\\right. T_{j} \\left(\\right. I \\left.\\right) \\left.\\right)\n$$\n\n(1)\n\nwhere $m \\in \\left{\\right. 1 , \\ldots , M \\left.\\right}$ indexes the base detector and $j \\in \\left{\\right. 1 , \\ldots , K \\left.\\right}$ indexes the TTA view. The set $\\mathcal{B}_{m , j}$ consists of the bounding-box coordinates and confidence scores generated by detector $m$ from the $j$-th transformed image:\n\n$$\n\\mathcal{B}_{m , j} = \\left{\\right. \\left(\\right. x_{1 , n} , y_{1 , n} , x_{2 , n} , y_{2 , n} , s_{n} \\left.\\right) \\left.\\right}_{n = 1}^{N_{m , j}}\n$$\n\n(2)\n\nwhere ($x_{1 , n} , y_{1 , n} , x_{2 , n} , y_{2 , n} , s_{n} \\left.\\right)$ denotes the coordinates of the $n$-th predicted bounding box, $s_{n}$ denotes its confidence score, and $N_{m , j}$ denotes the number of predictions generated by detector $m$ under TTA view $j$.\n\nTo fuse these predictions, their bounding-box coordinates are first mapped back to the coordinate system of the original image. Let $T_{j}^{- 1}$ denote the inverse transformation corresponding to $T_{j}$. The complete set of raw predictions generated by all base detectors and TTA views is expressed as\n\n$$\n\\mathcal{P}_{r a w} = \\cup_{m = 1}^{M} \\cup_{j = 1}^{K} T_{j}^{- 1} \\left(\\right. \\mathcal{B}_{m , j} \\left.\\right) , M = 3 , K = 6 .\n$$\n\n(3)\n\nHere, $T_{j}^{- 1}$ reverses the scaling and horizontal-flipping operations applied by $T_{j}$, thereby mapping the predicted bounding-box coordinates back to the original image space. Consequently, $\\mathcal{P}_{r a w}$ contains the predictions generated by all three base detectors across all six TTA views.\n\nAfter inverse coordinate mapping, the predictions from the six TTA views are first matched within each base detector according to class labels and a predefined IoU threshold. Specifically, the candidate boxes are sorted in descending order of confidence. The candidate with the highest current confidence is selected as the reference box, and its IoU with the remaining candidates of the same class is calculated. Candidates whose IoU with the reference box is not lower than the predefined TTA matching threshold are considered to correspond to the same target and are grouped together. The grouped candidates are then removed from the remaining candidate set, and the procedure is repeated until all candidate boxes have been assigned. Within each group, at most one candidate from each TTA view is retained. If multiple candidates from the same TTA view satisfy the matching criterion, only the one with the highest confidence score is retained.\n\nFor candidate boxes from different TTA views that are assigned to the same group, their bounding-box coordinates are computed as a confidence-weighted average. The confidence scores of the TTA views that actually contribute matched candidates are then averaged arithmetically to obtain a single candidate detection for the corresponding target from that base detector. After this within-model TTA matching and fusion procedure, YOLOv8n, YOLOv8s, and YOLOv8m each produce their respective model-specific candidate detections, which are subsequently used as the inputs to cross-model matching and WBF-based fusion.\n\nThis strategy enhances the scale adaptability of the detector without requiring model retraining. Specifically, enlarged views improve the effective pixel representation of distant small targets, and reduced views alleviate recognition instability in close-range large targets caused by local truncation or insufficient receptive fields. Horizontal flipping further provides prediction consistency under directional perturbations, thereby supplying more comprehensive candidate bounding boxes for the subsequent fusion stage. Importantly, the six TTA views of the same base detector are used only to construct model-specific candidate detections and are not treated as independent evidence sources in the subsequent Noisy-OR fusion. Therefore, each base detector contributes at most one model-level confidence score for a given candidate target.\n\n### 2.4. Spatial Consistency Fusion Based on WBF\n\nThe inputs to cross-model fusion are the model-specific candidate detections produced by YOLOv8n, YOLOv8s, and YOLOv8m after the within-model TTA matching and fusion procedure described in [Section 2.3](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F16\u002F1770#sec2dot3-agriculture-16-01770). To identify candidate detections from different base detectors that potentially correspond to the same target, all model-specific candidates are first sorted in descending order of confidence. The candidate with the highest current confidence is selected as the reference candidate, and its intersection over union (IoU) with the remaining candidates of the same class is calculated. Candidates whose IoU with the reference candidate is not lower than the predefined cross-model matching threshold are considered to correspond to the same target and are assigned to the same cross-model candidate cluster. The candidates assigned to the current cluster are then removed from the rem","MDPI Agriculture 16(16):1770 2026年8月18日","2026-08-18T00:00:00Z","论文",10,false,65,{"impact":17,"substance":18,"depth":19,"authority":17,"freshness":20,"relevant":21,"comment":22},12,18,16,7,1,"论文提出多模型融合框架提升芒果检测稳健性，对智慧农业细分领域有参考价值，但影响范围有限。",[24],{"name":10,"url":6},[26,27,28,29],"芒果检测","多模型融合","YOLOv8 MDPI Agriculture","果园",{"code":31,"slug":32,"name":33,"path":34,"parent_code":35},"B2","fruit","果树","\u002Ftopic\u002Fhorticulture\u002Ffruit","B",[],{"status":38,"locked":14,"version":39,"method":40,"evidence":41},"classified","2026-10-22.2","rules",{"candidates":42,"agricultural_evidence":45},[43],{"code":31,"score":44,"supported":45,"title":46,"abstract":49,"tags":50},19,true,[47,48],"芒果","orchard",[47],[],[52,53],"多模型融合 芒果检测 果园 YOLOv8 MDPI Agriculture","多模型融合 芒果检测","多模型融合芒果检测果园YOLOv8MDPIAgriculture-842",5,"2026-08-22T00:06:00.815340Z",{"total":58,"page":21,"page_size":58,"items":59},6,[60,110,192,233,304,367],{"id":61,"title":62,"url":63,"summary":64,"summary_zh":8,"content":65,"source_name":66,"source_url":8,"published_at":67,"category":12,"cover_url":8,"hotness":13,"is_selected":14,"score":68,"score_detail":69,"sources":74,"tags":76,"primary_topic":80,"secondary_topics":81,"classification":82,"search_phrases":100,"slug":103,"view_count":104,"doi":8,"paper":105,"created_at":109},1188,"A Feature Enhancement Framework for Joint Mango Fruit and Stem Detection in Complex Orchard Environments (复杂果园环境中芒果果实与果柄联合检测的特征增强框架)","https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F17\u002F1888","MDPI Agriculture 2026, 16(17), 1888, 2026年8月31日发表,作者来自华南农业大学工程学院和农业装备技术国家重点实验室。研究提出MangoNET,基于YOLOv11n的芒果果实与果柄联合检测框架,适用于复杂果园环境。P2高分辨率检测头保留小目标的精细空间信息,SPPF-ELAN聚合局部和上下文特征用于部分可见对象,SENet在光照变化下重新校准通道响应,WIoU v3调节具有不同定位质量的边界框样本。数据集包含1782张台农和金煌芒果原始图像,仅对训练集应用数据增强,将训练集大小从1172张增加到2886张图像。MangoNET果实和果柄F1分数分别为0.920和0.916,mAP50和mAP50-95值为0.941和0.690。","Next Article in Journal\n\n[Straw Biochar Incorporation Reduces CH 4 Emissions in Double-Cropping Rice Paddies and Partially Alleviates Warming-Induced Yield Losses](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F17\u002F1889)\n\n Previous Article in Journal\n\n[Tracking Insects in Controlled Experiments](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F17\u002F1887)\n\n Previous Article in Special Issue\n\n[ShuffleNetV2-hSimKD: A Lightweight Network for Plant Disease Detection](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F15\u002F1686)\n\n## Journals\n\n[Active Journals](https:\u002F\u002Fwww.mdpi.com\u002Fabout\u002Fjournals)[Find a Journal](https:\u002F\u002Fwww.mdpi.com\u002Fabout\u002Fjournalfinder)[Journal Proposal](https:\u002F\u002Fwww.mdpi.com\u002Fabout\u002Fjournals\u002Fproposal)[Proceedings Series](https:\u002F\u002Fwww.mdpi.com\u002Fabout\u002Fproceedings)\n\n## [Topics](https:\u002F\u002Fwww.mdpi.com\u002Ftopics)\n\n## Information\n\n[For Authors](https:\u002F\u002Fwww.mdpi.com\u002Fauthors)[For Reviewers](https:\u002F\u002Fwww.mdpi.com\u002Freviewers)[For Editors](https:\u002F\u002Fwww.mdpi.com\u002Feditors)[For Librarians](https:\u002F\u002Fwww.mdpi.com\u002Flibrarians)[For Publishers](https:\u002F\u002Fwww.mdpi.com\u002Fpublishing_services)[For Societies](https:\u002F\u002Fwww.mdpi.com\u002Fsocieties)[For Conference Organizers](https:\u002F\u002Fwww.mdpi.com\u002Fconference_organizers)\n\n[Open Access Policy](https:\u002F\u002Fwww.mdpi.com\u002Fopenaccess)[Institutional Open Access Program](https:\u002F\u002Fwww.mdpi.com\u002Fioap)[Special Issues Guidelines](https:\u002F\u002Fwww.mdpi.com\u002Fspecial_issues_guidelines)[Editorial Process](https:\u002F\u002Fwww.mdpi.com\u002Feditorial_process)[Research and Publication Ethics](https:\u002F\u002Fwww.mdpi.com\u002Fethics)[Article Processing Charge](https:\u002F\u002Fwww.mdpi.com\u002Fapc)[Awards](https:\u002F\u002Fwww.mdpi.com\u002Fawards)[Testimonials](https:\u002F\u002Fwww.mdpi.com\u002Ftestimonials)\n\n## [Author Services](https:\u002F\u002Fwww.mdpi.com\u002Fauthors\u002Fservices)\n\n## Initiatives\n\n[Sciforum](https:\u002F\u002Fsciforum.net\u002F)[MDPI Books](https:\u002F\u002Fwww.mdpi.com\u002Fbooks)[Preprints.org](https:\u002F\u002Fwww.preprints.org\u002F)[Scilit](https:\u002F\u002Fwww.scilit.com\u002F)[SciProfiles](https:\u002F\u002Fsciprofiles.com\u002F)[Encyclopedia](https:\u002F\u002Fencyclopedia.pub\u002F)[JAMS](https:\u002F\u002Fjams.pub\u002F)[Proceedings Series](https:\u002F\u002Fwww.mdpi.com\u002Fabout\u002Fproceedings)\n\n## About\n\n[Overview](https:\u002F\u002Fwww.mdpi.com\u002Fabout)[Contact](https:\u002F\u002Fwww.mdpi.com\u002Fabout\u002Fcontact)[Careers](https:\u002F\u002Fcareers.mdpi.com\u002F)[News](https:\u002F\u002Fwww.mdpi.com\u002Fabout\u002Fannouncements)[Press](https:\u002F\u002Fwww.mdpi.com\u002Fabout\u002Fpress)[Blog](http:\u002F\u002Fblog.mdpi.com\u002F)\n\n[Sign In \u002F Sign Up](https:\u002F\u002Fwww.mdpi.com\u002Fuser\u002Flogin)\n\n## Notice\n\n You can make submissions to other journals [here](https:\u002F\u002Fsusy.mdpi.com\u002Fuser\u002Fmanuscripts\u002Fupload). \n\n[_clear_](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F17\u002F1888)\n\n## Notice\n\nYou are accessing a machine-readable page. In order to be human-readable, please install an RSS reader.\n\n[Continue](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F17\u002F1888)[Cancel](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F17\u002F1888)\n\n[_clear_](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F17\u002F1888)\n\nAll articles published by MDPI are made immediately available worldwide under an open access license. No special permission is required to reuse all or part of the article published by MDPI, including figures and tables. For articles published under an open access Creative Common CC BY license, any part of the article may be reused without permission provided that the original article is clearly cited. For more information, please refer to [https:\u002F\u002Fwww.mdpi.com\u002Fopenaccess](https:\u002F\u002Fwww.mdpi.com\u002Fopenaccess).\n\nFeature papers represent the most advanced research with significant potential for high impact in the field. A Feature Paper should be a substantial original Article that involves several techniques or approaches, provides an outlook for future research directions and describes possible research applications.\n\nFeature papers are submitted upon individual invitation or recommendation by the scientific editors and must receive positive feedback from the reviewers.\n\nEditor’s Choice articles are based on recommendations by the scientific editors of MDPI journals from around the world. Editors select a small number of articles recently published in the journal that they believe will be particularly interesting to readers, or important in the respective research area. The aim is to provide a snapshot of some of the most exciting work published in the various research areas of the journal.\n\nOriginal Submission Date Received: .\n\n[![Image 5: MDPI Open Access Journals](https:\u002F\u002Fpub.mdpi-res.com\u002Fimg\u002Fdesign\u002Fmdpi-pub-logo-black-small1.svg?da3a8dcae975a41c?1788181901)](https:\u002F\u002Fwww.mdpi.com\u002F)\n\n[_clear_](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F17\u002F1888#)[_zoom\\_out\\_map_](https:\u002F\u002Fwww.mdpi.com\u002Ftoggle_desktop_layout_cookie \"Toggle desktop layout\")[_search_](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F17\u002F1888#)[_menu_](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F17\u002F1888# \"MDPI main page\")\n\n[![Image 6: MDPI Open Access Journals](https:\u002F\u002Fpub.mdpi-res.com\u002Fimg\u002Fdesign\u002Fmdpi-pub-logo-black-small1.svg?da3a8dcae975a41c?1788181901)](https:\u002F\u002Fwww.mdpi.com\u002F)\n\n*   [Journals](https:\u002F\u002Fwww.mdpi.com\u002Fabout\u002Fjournals)\n    *   \n        *   [Active Journals](https:\u002F\u002Fwww.mdpi.com\u002Fabout\u002Fjournals)\n        *   [Find a Journal](https:\u002F\u002Fwww.mdpi.com\u002Fabout\u002Fjournalfinder)\n        *   [Journal Proposal](https:\u002F\u002Fwww.mdpi.com\u002Fabout\u002Fjournals\u002Fproposal)\n        *   [Proceedings Series](https:\u002F\u002Fwww.mdpi.com\u002Fabout\u002Fproceedings)\n\n*   [Topics](https:\u002F\u002Fwww.mdpi.com\u002Ftopics)\n*   [Information](https:\u002F\u002Fwww.mdpi.com\u002Fauthors)\n    *   \n        *   [For Authors](https:\u002F\u002Fwww.mdpi.com\u002Fauthors)\n        *   [For Reviewers](https:\u002F\u002Fwww.mdpi.com\u002Freviewers)\n        *   [For Editors](https:\u002F\u002Fwww.mdpi.com\u002Feditors)\n        *   [For Librarians](https:\u002F\u002Fwww.mdpi.com\u002Flibrarians)\n        *   [For Publishers](https:\u002F\u002Fwww.mdpi.com\u002Fpublishing_services)\n        *   [For Societies](https:\u002F\u002Fwww.mdpi.com\u002Fsocieties)\n        *   [For Conference Organizers](https:\u002F\u002Fwww.mdpi.com\u002Fconference_organizers)\n\n        *   [Open Access Policy](https:\u002F\u002Fwww.mdpi.com\u002Fopenaccess)\n        *   [Institutional Open Access Program](https:\u002F\u002Fwww.mdpi.com\u002Fioap)\n        *   [Special Issues Guidelines](https:\u002F\u002Fwww.mdpi.com\u002Fspecial_issues_guidelines)\n        *   [Editorial Process](https:\u002F\u002Fwww.mdpi.com\u002Feditorial_process)\n        *   [Research and Publication Ethics](https:\u002F\u002Fwww.mdpi.com\u002Fethics)\n        *   [Article Processing Charge](https:\u002F\u002Fwww.mdpi.com\u002Fapc)\n        *   [Awards](https:\u002F\u002Fwww.mdpi.com\u002Fawards)\n        *   [Testimonials](https:\u002F\u002Fwww.mdpi.com\u002Ftestimonials)\n\n*   [Author Services](https:\u002F\u002Fwww.mdpi.com\u002Fauthors\u002Fservices)\n*   [Initiatives](https:\u002F\u002Fwww.mdpi.com\u002Fabout\u002Finitiatives)\n    *   \n        *   [Sciforum](https:\u002F\u002Fsciforum.net\u002F)\n        *   [MDPI Books](https:\u002F\u002Fwww.mdpi.com\u002Fbooks)\n        *   [Preprints.org](https:\u002F\u002Fwww.preprints.org\u002F)\n        *   [Scilit](https:\u002F\u002Fwww.scilit.com\u002F)\n        *   [SciProfiles](https:\u002F\u002Fsciprofiles.com\u002F)\n        *   [Encyclopedia](https:\u002F\u002Fencyclopedia.pub\u002F)\n        *   [JAMS](https:\u002F\u002Fjams.pub\u002F)\n        *   [Proceedings Series](https:\u002F\u002Fwww.mdpi.com\u002Fabout\u002Fproceedings)\n\n*   [About](https:\u002F\u002Fwww.mdpi.com\u002Fabout)\n    *   \n        *   [Overview](https:\u002F\u002Fwww.mdpi.com\u002Fabout)\n        *   [Contact](https:\u002F\u002Fwww.mdpi.com\u002Fabout\u002Fcontact)\n        *   [Careers](https:\u002F\u002Fcareers.mdpi.com\u002F)\n        *   [News](https:\u002F\u002Fwww.mdpi.com\u002Fabout\u002Fannouncements)\n        *   [Press](https:\u002F\u002Fwww.mdpi.com\u002Fabout\u002Fpress)\n        *   [Blog](http:\u002F\u002Fblog.mdpi.com\u002F)\n\n[Sign In \u002F Sign Up](https:\u002F\u002Fwww.mdpi.com\u002Fuser\u002Flogin)[Submit](https:\u002F\u002Fsusy.mdpi.com\u002Fuser\u002Fmanuscripts\u002Fupload?journal=agriculture)\n\nSearch for Articles:\n\nTitle \u002F Keyword\n\nAuthor \u002F Affiliation \u002F Email\n\nJournal\n\n[Agriculture](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F17\u002F1888)\n\nArticle Type\n\n[All Article Types](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F17\u002F1888)\n\n[Advanced Search](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F17\u002F1888#)\n\nSection\n\n[Select an Option](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F17\u002F1888)\n\nSpecial Issue\n\n[Select an Option](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F17\u002F1888)\n\nVolume\n\nIssue\n\nNumber\n\nPage\n\nLogical Operator Operator\n\nSearch Text\n\nSearch Type\n\n_add\\_circle\\_outline_\n\n_remove\\_circle\\_outline_\n\n[Journals](https:\u002F\u002Fwww.mdpi.com\u002Fabout\u002Fjournals)\n\n[Agriculture](https:\u002F\u002Fwww.mdpi.com\u002Fjournal\u002Fagriculture)\n\n[Volume 16](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16)\n\n[Issue 17](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F17)\n\n[10.3390\u002Fagriculture16171888](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F17\u002F1888#)\n\n[![Image 7: agriculture-logo](https:\u002F\u002Fpub.mdpi-res.com\u002Fimg\u002Fjournals\u002Fagriculture-logo.png?d9aa6395761dfdff)](https:\u002F\u002Fwww.mdpi.com\u002Fjournal\u002Fagriculture)\n\n[Submit to this Journal](https:\u002F\u002Fsusy.mdpi.com\u002Fuser\u002Fmanuscripts\u002Fupload?form%5Bjournal_id%5D%3D95)[Review for this Journal](https:\u002F\u002Fsusy.mdpi.com\u002Fvolunteer\u002Fjournals\u002Freview)[Propose a Special Issue](https:\u002F\u002Fwww.mdpi.com\u002Fjournalproposal\u002Fsendproposalspecialissue\u002Fagriculture)\n\n[►▼ Article Menu](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F17\u002F1888#)\n\n## Article Menu\n\n*   [Academic Editor](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F17\u002F1888#academic_editors)[![Image 8](https:\u002F\u002Fwww.mdpi.com\u002Fprofiles\u002F356481\u002Fthumb\u002FFrancesco_Marinello.jpg)Francesco Marinello](https:\u002F\u002Fsciprofiles.com\u002Fprofile\u002F356481?utm_source=mdpi.com&utm_medium=website&utm_campaign=avatar_name)   \n*   [Recommended Articles](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F17\u002F1888#)\n*   [Related Info Link](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F17\u002F1888#related)\n\n    *   [Google Scholar](http:\u002F\u002Fscholar.google.com\u002Fscholar?q=A%20Feature%20Enhancement%20Framework%20for%20Joint%20Mango%20Fruit%20and%20Stem%20Detection%20in%20Complex%20Orchard%20Environments)\n\n*   [More by Authors Links](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F17\u002F1888#authors)\n\n    *   [on DOAJ](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F17\u002F1888)\n    \n\n        *   [Lu, J.](http:\u002F\u002Fdoaj.org\u002Fsearch\u002Farticles?source=%7B%22query%22%3A%7B%22query_string%22%3A%7B%22query%22%3A%22%5C%22Jiahuan%20Lu%5C%22%22%2C%22default_operator%22%3A%22AND%22%2C%22default_field%22%3A%22bibjson.author.name%22%7D%7D%7D)\n\n        *   [Deng, Q.](http:\u002F\u002Fdoaj.org\u002Fsearch\u002Farticles?source=%7B%22query%22%3A%7B%22query_string%22%3A%7B%22query%22%3A%22%5C%22Qihan%20Deng%5C%22%22%2C%22default_operator%22%3A%22AND%22%2C%22default_field%22%3A%22bibjson.author.name%22%7D%7D%7D)\n\n        *   [Zheng, W.](http:\u002F\u002Fdoaj.org\u002Fsearch\u002Farticles?source=%7B%22query%22%3A%7B%22query_string%22%3A%7B%22query%22%3A%22%5C%22Weiping%20Zheng%5C%22%22%2C%22default_operator%22%3A%22AND%22%2C%22default_field%22%3A%22bibjson.author.name%22%7D%7D%7D)\n\n        *   [Cai, B.](http:\u002F\u002Fdoaj.org\u002Fsearch\u002Farticles?source=%7B%22query%22%3A%7B%22query_string%22%3A%7B%22query%22%3A%22%5C%22Binglong%20Cai%5C%22%22%2C%22default_operator%22%3A%22AND%22%2C%22default_field%22%3A%22bibjson.author.name%22%7D%7D%7D)\n\n        *   [Zeng, S.](http:\u002F\u002Fdoaj.org\u002Fsearch\u002Farticles?source=%7B%22query%22%3A%7B%22query_string%22%3A%7B%22query%22%3A%22%5C%22Shan%20Zeng%5C%22%22%2C%22default_operator%22%3A%22AND%22%2C%22default_field%22%3A%22bibjson.author.name%22%7D%7D%7D)\n\n        *   [Li, J.](http:\u002F\u002Fdoaj.org\u002Fsearch\u002Farticles?source=%7B%22query%22%3A%7B%22query_string%22%3A%7B%22query%22%3A%22%5C%22Jiehao%20Li%5C%22%22%2C%22default_operator%22%3A%22AND%22%2C%22default_field%22%3A%22bibjson.author.name%22%7D%7D%7D)\n\n    *   [on Google Scholar](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F17\u002F1888)\n    \n\n        *   [Lu, J.](http:\u002F\u002Fscholar.google.com\u002Fscholar?q=Jiahuan%20Lu)\n\n        *   [Deng, Q.](http:\u002F\u002Fscholar.google.com\u002Fscholar?q=Qihan%20Deng)\n\n        *   [Zheng, W.](http:\u002F\u002Fscholar.google.com\u002Fscholar?q=Weiping%20Zheng)\n\n        *   [Cai, B.](http:\u002F\u002Fscholar.google.com\u002Fscholar?q=Binglong%20Cai)\n\n        *   [Zeng, S.](http:\u002F\u002Fscholar.google.com\u002Fscholar?q=Shan%20Zeng)\n\n        *   [Li, J.](http:\u002F\u002Fscholar.google.com\u002Fscholar?q=Jiehao%20Li)\n\n    *   [on PubMed](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F17\u002F1888)\n    \n\n        *   [Lu, J.](http:\u002F\u002Fwww.pubmed.gov\u002F?cmd=Search&term=Jiahuan%20Lu)\n\n        *   [Deng, Q.](http:\u002F\u002Fwww.pubmed.gov\u002F?cmd=Search&term=Qihan%20Deng)\n\n        *   [Zheng, W.](http:\u002F\u002Fwww.pubmed.gov\u002F?cmd=Search&term=Weiping%20Zheng)\n\n        *   [Cai, B.](http:\u002F\u002Fwww.pubmed.gov\u002F?cmd=Search&term=Binglong%20Cai)\n\n        *   [Zeng, S.](http:\u002F\u002Fwww.pubmed.gov\u002F?cmd=Search&term=Shan%20Zeng)\n\n        *   [Li, J.](http:\u002F\u002Fwww.pubmed.gov\u002F?cmd=Search&term=Jiehao%20Li)\n\n[Article Views](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F17\u002F1888#metrics)\n\n[Citations-](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F17\u002F1888#metrics)\n\n*   [Table of Contents](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F17\u002F1888#table_of_contents)\n\n \n    *   [Abstract](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F17\u002F1888#)\n\nAltmetric[_share_ Share](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F17\u002F1888# \"Share\")[_announcement_ Help](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F17\u002F1888# \"Help\")[_format\\_quote_ Cite](javascript:void(0);)\n\n## Need Help?\n\n### Support\n\nFind support for a specific problem in the support section of our website.\n\n[Get Support](https:\u002F\u002Fwww.mdpi.com\u002Fabout\u002Fcontactform)\n\n### Feedback\n\nPlease let us know what you think of our products and services.\n\n[Give Feedback](https:\u002F\u002Fwww.mdpi.com\u002Ffeedback\u002Fsend)\n\n### Information\n\nVisit our dedicated information section to learn more about MDPI.\n\n[Get Information](https:\u002F\u002Fwww.mdpi.com\u002Fauthors)\n\n[_clear_](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F17\u002F1888)\n\n## JSmol Viewer\n\n[_clear_](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F17\u002F1888)\n\n_first\\_page_\n\n[Download PDF](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F17\u002F1888\u002Fpdf?version=1788183021)\n\n_settings_\n\n[Order Article Reprints](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F17\u002F1888\u002Freprints)\n\n Font 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\n**This is an early access version, the complete PDF, HTML, and XML versions will be available soon.**\n\nOpen Access Article\n\n# A Feature Enhancement Framework for Joint Mango Fruit and Stem Detection in Complex Orchard Environments\n\n by \n\n Jiahuan Lu\n\n![Image 9](https:\u002F\u002Fwww.mdpi.com\u002Fprofiles\u002F3614159\u002Fthumb\u002FJiahuan_Lu.png)Jiahuan Lu\n\n[SciProfiles](https:\u002F\u002Fsciprofiles.com\u002Fprofile\u002F3614159?utm_source=mdpi.com&utm_medium=website&utm_campaign=avatar_name)[Scilit](https:\u002F\u002Fscilit.com\u002Fscholars?q=Jiahuan%20Lu)[Preprints.org](https:\u002F\u002Fwww.preprints.org\u002Fsearch?condition_blocks=[{%22value%22:%22Jiahuan+Lu%22,%22type%22:%22author%22,%22operator%22:null}]&sort_field=relevance&sort_dir=desc&page=1&exact_match=true)[Google Scholar](https:\u002F\u002Fscholar.google.com\u002Fscholar?q=Jiahuan+Lu)\n\n 1,2[](mailto:jhlu@scau.edu.cn)[![Image 10: ORCID](https:\u002F\u002Fpub.mdpi-res.com\u002Fimg\u002Fdesign\u002Forcid.png?0465bc3812adeb52?1788181901)](https:\u002F\u002Forcid.org\u002F0009-0004-3223-6453), \n\n Qihan Deng\n\n![Image 11](https:\u002F\u002Fwww.mdpi.com\u002Fbundles\u002Fmdpisciprofileslink\u002Fimg\u002Funknown-user.png)Qihan Deng\n\n[SciProfiles](https:\u002F\u002Fsciprofiles.com\u002Fprofile\u002F5504892?utm_source=mdpi.com&utm_medium=website&utm_campaign=avatar_name)[Scilit](https:\u002F\u002Fscilit.com\u002Fscholars?q=Qihan%20Deng)[Preprints.org](https:\u002F\u002Fwww.preprints.org\u002Fsearch?condition_blocks=[{%22value%22:%22Qihan+Deng%22,%22type%22:%22author%22,%22operator%22:null}]&sort_field=relevance&sort_dir=desc&page=1&exact_match=true)[Google Scholar](https:\u002F\u002Fscholar.google.com\u002Fscholar?q=Qihan+Deng)\n\n 1,2, \n\n Weiping Zheng\n\n![Image 12](https:\u002F\u002Fwww.mdpi.com\u002Fbundles\u002Fmdpisciprofileslink\u002Fimg\u002Funknown-user.png)Weiping Zheng\n\n[SciProfiles](https:\u002F\u002Fsciprofiles.com\u002Fprofile\u002Fauthor\u002FVWh5MW9JVG9WbFhCRlFXN05Ybm5BUm9NVGI2QlFDZ1VVSkl2bmZiTjMzbz0=?utm_source=mdpi.com&utm_medium=website&utm_campaign=avatar_name)[Scilit](https:\u002F\u002Fscilit.com\u002Fscholars?q=Weiping%20Zheng)[Preprints.org](https:\u002F\u002Fwww.preprints.org\u002Fsearch?condition_blocks=[{%22value%22:%22Weiping+Zheng%22,%22type%22:%22author%22,%22operator%22:null}]&sort_field=relevance&sort_dir=desc&page=1&exact_match=true)[Google Scholar](https:\u002F\u002Fscholar.google.com\u002Fscholar?q=Weiping+Zheng)\n\n 1,2, \n\n Binglong Cai\n\n![Image 13](https:\u002F\u002Fwww.mdpi.com\u002Fbundles\u002Fmdpisciprofileslink\u002Fimg\u002Funknown-user.png)Binglong Cai\n\n[SciProfiles](https:\u002F\u002Fsciprofiles.com\u002Fprofile\u002Fauthor\u002FUEhjMTJLRVNvZktqTFc3cFV2VHBGNGYwaXRncVVHSEk0bU93YzhJQVptOD0=?utm_source=mdpi.com&utm_medium=website&utm_campaign=avatar_name)[Scilit](https:\u002F\u002Fscilit.com\u002Fscholars?q=Binglong%20Cai)[Preprints.org](https:\u002F\u002Fwww.preprints.org\u002Fsearch?condition_blocks=[{%22value%22:%22Binglong+Cai%22,%22type%22:%22author%22,%22operator%22:null}]&sort_field=relevance&sort_dir=desc&page=1&exact_match=true)[Google Scholar](https:\u002F\u002Fscholar.google.com\u002Fscholar?q=Binglong+Cai)\n\n 1,2, \n\n Shan Zeng\n\n![Image 14](https:\u002F\u002Fwww.mdpi.com\u002Fbundles\u002Fmdpisciprofileslink\u002Fimg\u002Funknown-user.png)Shan Zeng\n\n[SciProfiles](https:\u002F\u002Fsciprofiles.com\u002Fprofile\u002F2204687?utm_source=mdpi.com&utm_medium=website&utm_campaign=avatar_name)[Scilit](https:\u002F\u002Fscilit.com\u002Fscholars?q=Shan%20Zeng)[Preprints.org](https:\u002F\u002Fwww.preprints.org\u002Fsearch?condition_blocks=[{%22value%22:%22Shan+Zeng%22,%22type%22:%22author%22,%22operator%22:null}]&sort_field=relevance&sort_dir=desc&page=1&exact_match=true)[Google Scholar](https:\u002F\u002Fscholar.google.com\u002Fscholar?q=Shan+Zeng)\n\n 1,2[](mailto:shanzeng@scau.edu.cn)[![Image 15: ORCID](https:\u002F\u002Fpub.mdpi-res.com\u002Fimg\u002Fdesign\u002Forcid.png?0465bc3812adeb52?1788181901)](https:\u002F\u002Forcid.org\u002F0000-0002-2497-3363) and \n\n Jiehao Li\n\n![Image 16](https:\u002F\u002Fwww.mdpi.com\u002Fprofiles\u002F893272\u002Fthumb\u002FJiehao_Li.jpg)Jiehao Li\n\n[SciProfiles](https:\u002F\u002Fsciprofiles.com\u002Fprofile\u002F893272?utm_source=mdpi.com&utm_medium=website&utm_campaign=avatar_name)[Scilit](https:\u002F\u002Fscilit.com\u002Fscholars?q=Jiehao%20Li)[Preprints.org](https:\u002F\u002Fwww.preprints.org\u002Fsearch?condition_blocks=[{%22value%22:%22Jiehao+Li%22,%22type%22:%22author%22,%22operator%22:null}]&sort_field=relevance&sort_dir=desc&page=1&exact_match=true)[Google Scholar](https:\u002F\u002Fscholar.google.com\u002Fscholar?q=Jiehao+Li)\n\n 1,2,*[](mailto:jiehao.li@scau.edu.cn)[![Image 17: ORCID](https:\u002F\u002Fpub.mdpi-res.com\u002Fimg\u002Fdesign\u002Forcid.png?0465bc3812adeb52?1788181901)](https:\u002F\u002Forcid.org\u002F0000-0002-4946-4434)\n\n1\n\nCollege of Engineering, South China Agricultural University, Guangzhou 510642, China\n\n2\n\nNational Key Laboratory of Agricultural Equipment Technology, South China Agricultural University, Guangzhou 510642, China\n\n*\n\nAuthor to whom correspondence should be addressed. \n\n_Agriculture_**2026**, _16_(17), 1888; https:\u002F\u002Fdoi.org\u002F10.3390\u002Fagriculture16171888 (registering DOI) \n\nSubmission received: 30 July 2026 \u002F Revised: 27 August 2026 \u002F Accepted: 28 August 2026 \u002F Published: 31 August 2026\n\n (This article belongs to the Special Issue [Smart Sensor-Based Systems for Crop Monitoring](https:\u002F\u002Fwww.mdpi.com\u002Fjournal\u002Fagriculture\u002Fspecial_issues\u002F84MWNN99A6))\n\n[Download _keyboard\\_arrow\\_down_](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F17\u002F1888)\n\n[Download PDF](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F17\u002F1888\u002Fpdf?version=1788183021)\n\n[Download Supplementary Material](javascript:void(0);)\n\n[Versions Notes](https:\u002F\u002Fwww.mdpi.com\u002F2077-0472\u002F16\u002F17\u002F1888\u002Fnotes)\n\n## Abstract\n\n Reliable joint detection of mango fruits and stems is an essential upstream perception task for robotic harvesting, but remains challenging because stems are small, slender, frequently occluded, and visually degraded by illumination variation. This study proposes MangoNET, a YOLOv11n-based framework for joint mango fruit and stem detection in complex orchard environments. A P2 high-resolution detection head preserves fine spatial information for small targets, while SPPF-ELAN aggregates local and contextual features for partially visible objects. SENet recalibrates channel responses under illumination variation, and WIoU v3 regulates bounding-box samples with different localization qualities. A dataset containing 1782 original images of Tainong and Jinhuang mangoes was collected from two orchards and data augmentation was applied only to the training set, increasing its size from 1172 to 2886 images through rotation, contrast adjustment, and Gaussian noise addition. MangoNET achieved fruit and stem F1-scores of 0.920 and 0.916, respectively, with mAP50 and mAP50–95 values of 0.941 and 0.690. Compared with YOLOv11n, mAP50 and mAP50–95 increased by 1.6 and 2.9 percentage points, respectively, while stem recall increased from 0.877 to 0.906. Source-image-independent five-fold cross-validation yielded mean mAP50 and mAP50–95 values of 0.944 and 0.711, respectively. Pilot evaluations using images acquired by a UAV and an RGB-D camera in a geographically distinct orchard suggested that MangoNET could maintain detection performance in a different orchard environment. MangoNET s","Agriculture 2026, 16(17), 1888","2026-08-31T08:00:00Z",69,{"impact":70,"substance":71,"depth":18,"authority":17,"freshness":72,"relevant":21,"comment":73},15,20,4,"提出芒果果实与果柄联合检测的特征增强框架，方法新颖，数据详实，对农业机器人采摘有实际价值。",[75],{"name":66,"url":63},[77,78,79,26],"智慧农业","农业人工智能","目标检测",{"code":31,"slug":32,"name":33,"path":34,"parent_code":35},[],{"status":38,"locked":14,"version":39,"method":40,"evidence":83},{"candidates":84,"agricultural_evidence":45},[85,89,95],{"code":31,"score":44,"supported":45,"title":86,"abstract":87,"tags":88},[47,48],[47],[],{"code":90,"score":91,"supported":14,"title":92,"abstract":93,"tags":94},"I2",2,[],[],[78,79],{"code":96,"score":21,"supported":14,"title":97,"abstract":98,"tags":99},"I1",[],[],[77],[101,102],"农业人工智能 智慧农业 目标检测 芒果检测","农业人工智能 智慧农业","农业人工智能智慧农业目标检测芒果检测-1188",3,{"doi":8,"openalex_id":8,"authors":106,"venue":8,"cited_by_count":107,"oa_url":8,"card":8,"direction":8,"ingested_from":108},[],0,"agent","2026-09-01T00:03:40.048486Z",{"id":111,"title":112,"url":113,"summary":114,"summary_zh":115,"content":8,"source_name":116,"source_url":113,"published_at":117,"category":12,"cover_url":8,"hotness":13,"is_selected":14,"score":118,"score_detail":119,"sources":124,"tags":126,"primary_topic":132,"secondary_topics":133,"classification":140,"search_phrases":162,"slug":165,"view_count":107,"doi":166,"paper":167,"created_at":191},5547,"Apple quality and soil biological diversity in long-term organic and IPM apple orchards","https:\u002F\u002Fdoi.org\u002F10.3389\u002Ffsufs.2026.1943487","Introduction While several studies have compared fruit quality in organic and conventional apple production, integrated assessments of fruit quality and soil biological diversity remain limited. Majority of previous studies have been based on relatively short-term experimental comparisons, although responses of soil biota and soil properties often require many years of consistent management to become fully expressed. Methods This study compared yield, fruit quality, and soil biological diversity in long-term organic (ORG) apple orchard and long-term integrated pest management (IPM) orchard under Nordic conditions. The study was conducted in 2022–2024 at the Polli Horticultural Research Centre (Estonia) using the cultivars “Katre”, “Krista,” and “Alesja”. Apple, yield, fruit quality attributes (soluble solids and titratable acids content and their ratio, ascorbic acid content and total polyphenols, total anthocyanins, fruit firmness, fruit colour, fruit mineral elements content), soil chemical parameters, earthworm and springtail communities were analysed. Results and Discussion Higher total and fresh-market yields were observed in the IPM orchard in two of the three cultivars, whereas the proportion of fresh-market fruit did not differ consistently between the two orchards. When averaged across cultivars, apples from ORG orchard exhibited more intense peel coloration. Differences in taste-related traits and antioxidant composition between cultivation systems were cultivar- and year-dependent rather than consistently associated with cultivation system. Total polyphenol content was consistently higher under ORG management only in “Katre,” while “Alesja” exhibited the highest total polyphenol and anthocyanin contents irrespective of cultivation system, highlighting the dominant role of genotype. Soil in the ORG orchard had higher pH and greater phosphorus and calcium contents, supporting a higher proportion of anecic earthworms, particularly Aporrectodea longa . Total earthworm abundance was higher in the IPM orchard, springtail abundance did not differ significantly between systems. These results suggest that the conditions in the long-term ORG were more favourable for ecologically important earthworm groups without consistently affecting internal fruit quality compared to IPM orchard.","引言 尽管已有若干研究比较了有机与常规苹果生产中的果实品质，但结合果实品质与土壤生物多样性的综合评价仍较为有限。以往研究大多基于相对短期的试验比较，而土壤生物区系和土壤性质的响应往往需要多年持续一致的管理才能充分显现。方法 本研究在北欧条件下比较了长期有机（ORG）苹果园与长期综合病虫害管理（IPM）苹果园的产量、果实品质及土壤生物多样性。研究于2022—2024年在爱沙尼亚Polli园艺研究中心开展，供试品种为“Katre”“Krista”和“Alesja”。分析了苹果产量、果实品质指标（可溶性固形物与可滴定酸含量及其比值、抗坏血酸含量、总多酚、总花青素、果实硬度、果实色泽、果实矿质元素含量）、土壤化学参数、蚯蚓和跳虫群落。结果与讨论 在3个品种中的2个品种上，IPM果园的总产量和鲜食市场产量较高，而鲜食市场果实比例在两种果园间并无一致差异。按品种平均，ORG果园苹果果皮色泽更深。栽培系统间口感相关性状和抗氧化成分的差异因品种和年份而异，而非与栽培系统稳定相关。总多酚含量仅在“Katre”中于ORG管理下持续较高，而“Alesja”无论栽培系统如何均表现出最高的总多酚和花青素含量，凸显了基因型的主导作用。ORG果园土壤具有较高的pH以及较高的磷和钙含量，支持了较高比例的内栖型蚯蚓，尤其是长穴居蚯蚓（Aporrectodea longa）。IPM果园蚯蚓总丰度较高，跳虫丰度在系统间无显著差异。这些结果表明，与IPM果园相比，长期ORG的条件更有利于生态上重要的蚯蚓类群，但并未对果实内部品质产生一致影响。","Frontiers in Sustainable Food Systems","2026-10-21T00:00:00Z",66,{"impact":120,"substance":71,"depth":19,"authority":121,"freshness":122,"relevant":21,"comment":123},8,13,9,"北欧长期定位试验对比有机与IPM苹果园的果实品质与土壤生物多样性，结论有新意但属区域尺度研究，公共影响有限。",[125],{"name":116,"url":113},[127,128,129,130,131],"长期定位试验","有机果园","土壤生物多样性","苹果品质","IPM",{"code":31,"slug":32,"name":33,"path":34,"parent_code":35},[134],{"code":135,"slug":136,"name":137,"path":138,"parent_code":139},"H2","soil","土壤健康与改良","\u002Ftopic\u002Fenvironment\u002Fsoil","H",{"status":38,"locked":14,"version":39,"method":40,"evidence":141},{"candidates":142,"agricultural_evidence":45},[143,150,156],{"code":31,"score":144,"supported":45,"title":145,"abstract":147,"tags":149},17,[146],"apple",[148,48,146],"苹果",[],{"code":135,"score":151,"supported":45,"title":152,"abstract":153,"tags":155},14,[136],[154,136],"土壤",[],{"code":157,"score":104,"supported":45,"title":158,"abstract":159,"tags":161},"H6",[],[160],"生物多样性",[],[163,164],"爱沙尼亚 Polli 苹果 有机 IPM","苹果园 蚯蚓 跳虫 土壤","爱沙尼亚Polli苹果有机IPM-5547","10.3389\u002Ffsufs.2026.1943487",{"doi":166,"openalex_id":168,"authors":169,"venue":116,"cited_by_count":107,"oa_url":113,"card":184,"direction":188,"ingested_from":190},"W7223104118",[170,173,176,178,181],{"name":171,"orcid":172},"Ulvi Moor","https:\u002F\u002Forcid.org\u002F0000-0002-8808-8494",{"name":174,"orcid":175},"Annely Kuu","https:\u002F\u002Forcid.org\u002F0000-0002-0335-6243",{"name":177,"orcid":8},"K. Kahu",{"name":179,"orcid":180},"P. Põldma","https:\u002F\u002Forcid.org\u002F0000-0002-7906-823X",{"name":182,"orcid":183},"Reelika Rätsep","https:\u002F\u002Forcid.org\u002F0000-0002-8673-9978",{"tldr":185,"method":186,"finding":187,"direction":188,"opportunity":189},"比较长期有机与IPM苹果园的果实品质和土壤生物多样性。","在爱沙尼亚长期试验园测定三个品种的产量、品质及蚯蚓和跳虫群落。","有机园果皮色泽更深、土壤pH和磷钙更高，但果实内在品质差异多随品种和年份变化。","农业绿色发展与碳","可结合长期定位试验，量化土壤动物群落与果实品质的关联机制，并拓展多区域验证。","openalex","2026-10-22T11:20:15.557635Z",{"id":193,"title":194,"url":195,"summary":196,"summary_zh":8,"content":197,"source_name":198,"source_url":8,"published_at":199,"category":200,"cover_url":8,"hotness":13,"is_selected":14,"score":15,"score_detail":201,"sources":203,"tags":205,"primary_topic":210,"secondary_topics":211,"classification":212,"search_phrases":228,"slug":231,"view_count":107,"doi":8,"paper":8,"created_at":232},5516,"上海农科院葡萄团队赴银川参加第47届世界葡萄与葡萄酒大会","https:\u002F\u002Fwww.saas.sh.cn\u002Fhzjl\u002Fgjhz\u002Fcontent_57128","10月12日至16日第47届世界葡萄与葡萄酒大会在银川举办，智慧农业研究中心葡萄团队负责人奚晓军研究员作专题报告，阐明疏穗与环剥处理对果实糖、花青素及脱落酸积累的源库调控机制。团队还观摩酿酒葡萄AI综合作业机器人成果展示，与研发单位围绕葡萄智慧装备与果园智能化管理深入交流。","2026年10月12-16日，第47届世界葡萄与葡萄酒大会在银川举办，智慧农业研究中心葡萄团队负责人奚晓军研究员一行赴会交流。团队围绕“提升葡萄种植生产韧性与可持续发展能力”主题，向全球同行展示最新科研成果。\n\n会上，奚晓军作“Changes in sugar, anthocyanin and ABA accumulation in response to cluster thinning and girdling in ‘Jumeigui’grape”专题报告。报告围绕葡萄疏穗与环剥处理开展研究，阐明源库比对果实糖、花青素及脱落酸积累的调控机制，为葡萄优质栽培技术提供理论支撑，报告内容引发国内外园艺专家广泛研讨。参会期间，团队前往银川国际会展中心观摩酿酒葡萄 AI 综合作业机器人成果展示，详细了解该装备在葡萄园智能作业方面的技术特点与应用前景，并与研发单位围绕葡萄智慧装备研发、果园智能化管理等方向开展深入交流。\n\n本次参会是葡萄团队持续推进科研国际化、对标国际前沿开展研究的重要实践。","上海市农业科学院 2026-10-19","2026-10-19T00:00:00Z","报道",{"impact":19,"substance":144,"depth":121,"authority":121,"freshness":58,"relevant":21,"comment":202},"国际学术会议交流并涉及葡萄AI作业机器人观摩，有实质科研内容与行业参考价值，但属团队参会动态，影响面有限。",[204],{"name":198,"url":195},[77,206,207,208,209],"智能农机","葡萄","栽培技术","国际交流",{"code":31,"slug":32,"name":33,"path":34,"parent_code":35},[],{"status":38,"locked":14,"version":39,"method":40,"evidence":213},{"candidates":214,"agricultural_evidence":45},[215,219,223],{"code":31,"score":17,"supported":45,"title":216,"abstract":217,"tags":218},[207],[207],[207],{"code":96,"score":72,"supported":45,"title":220,"abstract":221,"tags":222},[],[77],[77],{"code":224,"score":21,"supported":14,"title":225,"abstract":226,"tags":227},"F5",[],[],[206],[229,230],"上海农科院 葡萄 银川","世界葡萄与葡萄酒大会 奚晓军","上海农科院葡萄银川-5516","2026-10-22T00:07:04.061911Z",{"id":234,"title":235,"url":236,"summary":237,"summary_zh":238,"content":8,"source_name":116,"source_url":236,"published_at":239,"category":12,"cover_url":8,"hotness":13,"is_selected":14,"score":240,"score_detail":241,"sources":244,"tags":246,"primary_topic":252,"secondary_topics":253,"classification":254,"search_phrases":273,"slug":276,"view_count":107,"doi":277,"paper":278,"created_at":303},5468,"Farm-level technical efficiency in temperate horticulture: evidence from apple, pear, and cherry growers","https:\u002F\u002Fdoi.org\u002F10.3389\u002Ffsufs.2026.1928386","Improving resource-use efficiency is fundamental to enhancing agricultural productivity and promoting sustainable horticultural development, particularly in regions where opportunities for expanding cultivated land and increasing input use are limited. This study evaluates the technical efficiency of apple, pear, and cherry growers in the Kashmir Valley and examines the socioeconomic, institutional, and regional factors associated with farm-level technical efficiency. Primary data were collected from 418 horticultural households during 2023 using a multi-stage sampling approach. Technical efficiency was first estimated separately for apple, pear, and cherry growers using an input-oriented Data Envelopment Analysis (DEA) framework under both Constant Returns to Scale (CRS) and Variable Returns to Scale (VRS) assumptions. Scale efficiency was subsequently estimated, while a pooled DEA model was employed to generate farm-level technical efficiency scores for the second-stage analysis. The determinants of technical efficiency were examined using Tobit regression, supported by a fractional logit model as a robustness check, and regional disparities were assessed using Kruskal–Wallis and Mann–Whitney non-parametric tests. The results indicate that apple growers achieved the highest overall technical efficiency, followed by pear and cherry growers. The VRS estimates reveal substantially higher levels of pure technical efficiency, whereas the scale efficiency analysis suggests that a considerable proportion of technical inefficiency arises from farms operating below their optimal production scale. The second-stage analysis identifies family size, education, irrigation facilities, access to credit, and cold storage infrastructure as significant positive factors associated with farm-level technical efficiency, while significant regional disparities are observed across the Kashmir Valley. The robustness analysis confirms the consistency of the estimated relationships. These findings highlight that improving horticultural performance requires not only efficient farm-level resource management but also stronger institutional support, investments in irrigation and post-harvest infrastructure, improved access to rural credit, and region-specific development strategies. The study contributes to the literature by integrating crop-specific and pooled efficiency analyses with socioeconomic and regional assessments, providing policy-relevant evidence for promoting resource-efficient and sustainable horticultural production.","提高资源利用效率是提升农业生产力、促进园艺产业可持续发展的基础，在耕地扩张和投入增加机会有限的地区尤为如此。本研究评估了克什米尔谷地苹果、梨和樱桃种植者的技术效率，并考察了与农场层面技术效率相关的社会经济、制度和区域因素。研究采用多阶段抽样方法，于2023年从418个园艺农户收集了一手数据。首先，在规模报酬不变（CRS）和规模报酬可变（VRS）两种假设下，运用投入导向的数据包络分析（DEA）框架分别估算苹果、梨和樱桃种植者的技术效率。随后估算规模效率，并采用混合DEA模型生成农场层面技术效率得分用于第二阶段分析。技术效率的决定因素通过Tobit回归进行检验，并以分数logit模型作为稳健性检验加以支持，区域差异则采用Kruskal–Wallis和Mann–Whitney非参数检验进行评估。结果表明，苹果种植者总体技术效率最高，梨和樱桃种植者次之。VRS估计显示纯技术效率水平显著更高，而规模效率分析表明，相当比例的技术无效率源于农场经营规模低于最优生产规模。第二阶段分析识别出家庭规模、教育、灌溉设施、信贷获取和冷库基础设施是与农场层面技术效率显著正相关的因素，同时在克什米尔谷地观察到显著的区域差异。稳健性分析证实了所估计关系的一致性。这些发现表明，提升园艺产业绩效不仅需要有效的农场层面资源管理，还需要更强的制度支持、对灌溉和采后基础设施的投资、改善农村信贷获取以及因地制宜的发展策略。本研究通过将作物特定效率分析与混合效率分析同社会经济和区域评估相结合，为促进资源利用效率提升提供了具有政策相关性的证据。","2026-10-20T00:00:00Z",73,{"impact":17,"substance":242,"depth":18,"authority":121,"freshness":122,"relevant":21,"comment":243},21,"基于418户果农一手数据，用DEA与Tobit回归分析技术效率，方法规范、结论有政策参考价值，但属区域性研究，影响范围有限。",[245],{"name":116,"url":236},[247,248,249,250,251],"可持续农业","园艺产业","苹果种植","技术效率","数据包络分析",{"code":31,"slug":32,"name":33,"path":34,"parent_code":35},[],{"status":38,"locked":14,"version":39,"method":40,"evidence":255},{"candidates":256,"agricultural_evidence":45},[257,261,268],{"code":31,"score":151,"supported":45,"title":258,"abstract":259,"tags":260},[146],[148,146],[],{"code":262,"score":58,"supported":45,"title":263,"abstract":264,"tags":267},"H3",[],[265,266],"灌溉","irrigation",[],{"code":269,"score":21,"supported":14,"title":270,"abstract":271,"tags":272},"H8",[],[],[247],[274,275],"数据包络分析 可持续农业 园艺产业 技术效率","数据包络分析 可持续农业","数据包络分析可持续农业园艺产业技术效率-5468","10.3389\u002Ffsufs.2026.1928386",{"doi":277,"openalex_id":279,"authors":280,"venue":116,"cited_by_count":107,"oa_url":236,"card":297,"direction":301,"ingested_from":190},"W7223025732",[281,283,286,289,292,295],{"name":282,"orcid":8},"Seerat Sajjad",{"name":284,"orcid":285},"Khursheed Hussain Dar","https:\u002F\u002Forcid.org\u002F0000-0002-3925-0758",{"name":287,"orcid":288},"Anam Ul Haq Ganie","https:\u002F\u002Forcid.org\u002F0000-0002-4404-5011",{"name":290,"orcid":291},"Ghadda Mohamed Awad Yousif","https:\u002F\u002Forcid.org\u002F0000-0003-4183-2957",{"name":293,"orcid":294},"Mohd Salim","https:\u002F\u002Forcid.org\u002F0000-0003-0162-8433",{"name":296,"orcid":8},"Mahdy S. Othman",{"tldr":298,"method":299,"finding":300,"direction":301,"opportunity":302},"评估克什米尔谷地苹果、梨、樱桃种植户的技术效率及其影响因素。","418户调查数据，DEA测效率，Tobit回归与分数logit分析影响因素。","苹果效率最高，规模效率不足是主因；灌溉、信贷、冷库等显著提升效率。","数字乡村与农业信息化","可结合遥感与物联网数据构建果园投入产出实时效率监测，弥补自报数据局限。","2026-10-21T23:30:07.145395Z",{"id":305,"title":306,"url":307,"summary":308,"summary_zh":309,"content":8,"source_name":116,"source_url":307,"published_at":239,"category":12,"cover_url":8,"hotness":13,"is_selected":14,"score":310,"score_detail":311,"sources":314,"tags":316,"primary_topic":321,"secondary_topics":322,"classification":323,"search_phrases":347,"slug":350,"view_count":107,"doi":351,"paper":352,"created_at":366},5464,"From symbolic to substantive: How subsidy design shapes cooperative roles and high-quality rural industrial development?","https:\u002F\u002Fdoi.org\u002F10.3389\u002Ffsufs.2026.1949418","Introduction Government subsidies do not necessarily produce high-quality rural industrial development (HQRID), as different subsidy designs elicit different behavioral responses from cooperatives and smallholders. This study examines how subsidy design shapes the strategies of local governments, cooperatives, and farmers. Methods Drawing on the Dingzhuang Grape Cooperative in Jurong, China, this study builds a tripartite evolutionary game model to examine how creation support and function support shape the strategies of the local government, cooperative, and farmer. Results The results distinguish two contrasting pathways. Creation support tends to induce symbolic compliance, in which cooperatives adopt formal structures without providing training, quality control, or market access, leading the system toward a low-quality equilibrium. Function support aligns cooperative incentives with verifiable performance and encourages substantive services that connect smallholders to higher-value markets. Three conclusions follow. First, the transition to substantive service requires function support that gives cooperatives sufficiently strong incentives. Second, substantive services lower the quality-premium share that farmers require, strengthening the economic case for high-quality production. Third, function support is bounded by fiscal capacity, and excessive performance-linked payments can reverse the transition. Discussion Policy should therefore tie support to measurable gains in farmer income and product quality. The study contributes to induced institutional change theory by specifying that external inducements produce substantive institutions only when rewards are tied to verifiable performance rather than to registration, a boundary condition that explains why supported organizations may remain hollow.","引言 政府补贴并不必然带来高质量的农村产业发展（HQRID），因为不同的补贴设计会引发合作社与小农户不同的行为反应。本研究考察补贴设计如何塑造地方政府、合作社与农户的策略。方法 本研究以中国句容市丁庄葡萄合作社为案例，构建三方演化博弈模型，考察创建支持与功能支持如何塑造地方政府、合作社与农户的策略。结果 结果区分出两条截然不同的路径。创建支持倾向于诱发象征性遵从，即合作社采取正式结构却不提供培训、质量控制或市场准入，使系统趋向低质量均衡。功能支持使合作社激励与可验证绩效相一致，并鼓励将小农户连接至更高价值市场的实质性服务。由此得出三点结论。第一，向实质性服务的转变需要功能支持，使合作社获得足够强的激励。第二，实质性服务降低了农户所需的质量溢价份额，强化了高质量生产的经济合理性。第三，功能支持受财政能力约束，过度的绩效挂钩支付可能逆转这一转变。讨论 因此，政策应将支持与农户收入和产品质量的可测量改善挂钩。本研究对诱致性制度变迁理论作出贡献，明确指出外部诱因只有在奖励与可验证绩效而非与登记注册挂钩时才会产生实质性制度，这一边界条件解释了受支持组织为何可能保持空壳化。",77,{"impact":19,"substance":312,"depth":18,"authority":121,"freshness":120,"relevant":21,"comment":313},22,"以句容葡萄合作社为案例的三方演化博弈研究，区分创建型与功能型补贴对合作社行为的影响，结论对补贴政策设计有实质参考价值。",[315],{"name":116,"url":307},[317,318,319,320],"乡村产业","合作社","小农户","补贴设计",{"code":31,"slug":32,"name":33,"path":34,"parent_code":35},[],{"status":38,"locked":14,"version":39,"method":40,"evidence":324},{"candidates":325,"agricultural_evidence":45},[326,331,336,342],{"code":31,"score":58,"supported":45,"title":327,"abstract":328,"tags":330},[],[207,329],"grape",[],{"code":332,"score":55,"supported":45,"title":333,"abstract":334,"tags":335},"G2",[],[319],[318,319],{"code":337,"score":104,"supported":45,"title":338,"abstract":339,"tags":341},"G6",[],[340],"农户收入",[],{"code":343,"score":21,"supported":14,"title":344,"abstract":345,"tags":346},"G4",[],[],[317],[348,349],"句容 丁庄 葡萄合作社","补贴设计 合作社 小农户","句容丁庄葡萄合作社-5464","10.3389\u002Ffsufs.2026.1949418",{"doi":351,"openalex_id":353,"authors":354,"venue":116,"cited_by_count":107,"oa_url":307,"card":361,"direction":301,"ingested_from":190},"W7222997429",[355,357,359],{"name":356,"orcid":8},"Man Huang",{"name":358,"orcid":8},"蒋礼文",{"name":360,"orcid":8},"Jincai Zhuang",{"tldr":362,"method":363,"finding":364,"direction":301,"opportunity":365},"用三方演化博弈分析补贴设计如何影响合作社行为与农村产业高质量发展。","构建政府、合作社、农户三方演化博弈模型，基于句容丁庄葡萄合作社案例。","创建型补贴诱发象征性合规，功能型补贴促成实质性服务并降低农户质量溢价要求。","可将补贴绩效挂钩机制与可验证数据采集结合，研究数字化考核如何抑制合作社空心化。","2026-10-21T23:30:06.732822Z",{"id":368,"title":369,"url":370,"summary":371,"summary_zh":372,"content":8,"source_name":373,"source_url":370,"published_at":374,"category":12,"cover_url":8,"hotness":13,"is_selected":14,"score":375,"score_detail":376,"sources":378,"tags":380,"primary_topic":384,"secondary_topics":385,"classification":386,"search_phrases":410,"slug":413,"view_count":107,"doi":414,"paper":415,"created_at":437},5401,"A HIERARCHICAL DISTRIBUTED EDGE AI FRAMEWORK FOR REAL-TIME APPLE LEAF DISEASE CLASSIFICATION","https:\u002F\u002Fdoi.org\u002F10.22190\u002Ffuacr260506002g","The application of artificial intelligence (AI) in agriculture has transformed crop management, particularly in the detection of plant diseases. Apple trees, a key agricultural commodity, are vulnerable to various diseases, which can severely impact crop yields. Traditional disease detection methods, often reliant on manual inspections, are time-consuming and prone to error. In response, this study proposes a novel two-stage edge AI framework for apple leaf disease classification that integrates RGB imagery with distributed edge computing. The first stage performs binary classification to distinguish between healthy and diseased leaves, achieving an F1 score of 95.50%. This optimization is achieved by utilizing lightweight models on edge devices, thereby enhancing computational efficiency. The second stage employs a consensus-based multiclass classification approach, achieving a 96.44% F1 score in identifying specific diseases. This dual-stage system, supported by an edge AI architecture, ensures real-time disease detection while minimizing resource consumption and enhancing scalability in large-scale agricultural environments. The full system achieved a 95.50% accuracy and F1 score on the special dataset. The results demonstrate the system’s efficacy in controlled laboratory conditions, offering a promising solution for precise, real-time agricultural monitoring. This study underscores the potential of distributed edge AI systems for scalable and efficient disease detection in precision agriculture.","人工智能（AI）在农业中的应用已深刻改变了作物管理方式，尤其是在植物病害检测方面。苹果树作为重要的农业经济作物，易受多种病害侵袭，严重影响作物产量。传统的病害检测方法通常依赖人工检查，耗时且易出错。为此，本研究提出了一种新颖的两阶段边缘AI框架，用于苹果叶片病害分类，该框架将RGB图像与分布式边缘计算相结合。第一阶段执行二分类以区分健康叶片与患病叶片，F1分数达到95.50%。该优化通过在边缘设备上采用轻量级模型实现，从而提升了计算效率。第二阶段采用基于共识的多类分类方法，在识别具体病害方面F1分数达到96.44%。该双阶段系统在边缘AI架构的支持下，实现了实时病害检测，同时最大限度地降低了资源消耗，并增强了在大规模农业环境中的可扩展性。完整系统在专用数据集上达到了95.50%的准确率和F1分数。结果表明，该系统在受控实验室条件下具有有效性，为精准、实时的农业监测提供了一种有前景的解决方案。本研究凸显了分布式边缘AI系统在精准农业中实现可扩展且高效病害检测的潜力。","Facta Universitatis Series Automatic Control and Robotics","2026-10-18T00:00:00Z",76,{"impact":18,"substance":242,"depth":144,"authority":17,"freshness":120,"relevant":21,"comment":377},"该研究提出两级边缘AI框架，在苹果叶片病害识别上取得95%以上F1值，方法新颖、数据可靠，对精准农业实时监测有参考价值，但尚处实验室验证阶段，产业影响有限。",[379],{"name":373,"url":370},[77,78,381,382,383],"边缘计算","作物病害识别","苹果病害",{"code":31,"slug":32,"name":33,"path":34,"parent_code":35},[],{"status":38,"locked":14,"version":39,"method":40,"evidence":387},{"candidates":388,"agricultural_evidence":45},[389,393,399,405],{"code":31,"score":18,"supported":45,"title":390,"abstract":391,"tags":392},[146],[33,148,146],[383],{"code":96,"score":20,"supported":45,"title":394,"abstract":395,"tags":398},[],[396,397],"精准农业","precision agriculture",[77],{"code":400,"score":20,"supported":45,"title":401,"abstract":402,"tags":404},"I5",[],[381,403],"edge computing",[381],{"code":90,"score":55,"supported":45,"title":406,"abstract":407,"tags":409},[],[408],"artificial intelligence",[382,78],[411,412],"苹果叶片病害 边缘AI","分布式边缘计算 病害分类","苹果叶片病害边缘AI-5401","10.22190\u002Ffuacr260506002g",{"doi":414,"openalex_id":416,"authors":417,"venue":373,"cited_by_count":107,"oa_url":370,"card":430,"direction":436,"ingested_from":190},"W7222964752",[418,421,424,427],{"name":419,"orcid":420},"Milan Grujev","https:\u002F\u002Forcid.org\u002F0009-0002-3211-6844",{"name":422,"orcid":423},"Dragan H. Stojanovic","https:\u002F\u002Forcid.org\u002F0000-0003-4893-3814",{"name":425,"orcid":426},"Miloš Ilić","https:\u002F\u002Forcid.org\u002F0000-0002-9696-2874",{"name":428,"orcid":429},"Aleksandar Milosavljević","https:\u002F\u002Forcid.org\u002F0000-0003-3057-7727",{"tldr":431,"method":432,"finding":433,"direction":434,"opportunity":435},"提出两级边缘AI框架，用RGB图像实时分类苹果叶片病害。","分布式边缘计算+轻量模型，先二分类再共识多分类。","两级系统准确率与F1均达95.50%，多分类F1为96.44%。","智慧农业 \u002F 农业物联网","仅在受控实验室验证，可延伸至田间复杂光照与多作物泛化及边缘能耗优化。","农业人工智能与决策模型","2026-10-20T23:30:33.830748Z"]