[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"item-467":3,"related-467":67},{"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":61,"slug":64,"view_count":65,"doi":8,"paper":8,"created_at":66},467,"Advances in Binocular Stereo Vision-Driven 3D Perception and Intelligent Analysis Methods for Agriculture","https:\u002F\u002Fwww.mdpi.com\u002F2076-3417\u002F16\u002F16\u002F7957","本文系统综述双目立体视觉在农业应用中的最新进展,涵盖系统架构、传统与基于深度学习的立体匹配方法、点云重建技术及新兴的3D高斯溅射等监督策略。综述总结了关键应用,包括高通量表型分析、果实定位与机器人采摘、杂草检测与精准喷施、自主导航以及畜禽体况评估,突出其在多任务农业感知系统中的作用。论文还讨论了低纹理匹配困难、复杂环境中的遮挡、跨域泛化、实时轻量化部署及有限数据集可用性等主要挑战,并展望基于基础模型的视觉感知、自监督与弱监督学习、多模态融合及边缘高效模型设计的未来方向。原文标题:Advances in Binocular Stereo Vision-Driven 3D Perception and Intelligent Analysis Methods for Agriculture,作者:江苏大学Ye R、Wang J、Kong Z、Ou M,发表于MDPI Applied Sciences 2026年第16卷第16期7957。",null,"Logical Operator Operator\n\nSearch Text\n\nSearch Type\n\n_add\\_circle\\_outline_\n\n_remove\\_circle\\_outline_\n\n[![Image 1: applsci-logo](https:\u002F\u002Fpub.mdpi-res.com\u002Fimg\u002Fjournals\u002Fapplsci-logo.png?c706cab522490ca4)](https:\u002F\u002Fwww.mdpi.com\u002Fjournal\u002Fapplsci)\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 Review\n\nby \nRui Ye\n\n 1,2, \nJialin Wang\n\n 1,2[](mailto:2222516041@stmail.ujs.edu.cn), \nZhihao Kong\n\n 1,2[](mailto:2212516020@stmail.ujs.edu.cn) and \nMingxiong Ou\n\n 1,2,*[](mailto:myomx@ujs.edu.cn)\n\n1\n\nSchool of Agricultural Engineering, Jiangsu University, Zhenjiang 212013, China\n\n2\n\nHigh-Tech Key Laboratory of Agricultural Equipment and Intelligence of Jiangsu Province, Jiangsu University, Zhenjiang 212013, China\n\n*\n\nAuthor to whom correspondence should be addressed.\n\nSubmission received: 3 July 2026 \u002F Revised: 4 August 2026 \u002F Accepted: 8 August 2026 \u002F Published: 10 August 2026\n\n## Abstract\n\nBinocular stereo vision is a low-cost and scalable 3D perception technology that shows strong potential in agricultural phenotyping and smart agriculture. By estimating depth from multi-view RGB images, it enables non-contact, high-precision sensing of crop structure, canopy morphology, growth dynamics, and livestock traits, providing essential support for digital and intelligent agricultural production. With recent advances in deep learning-based stereo matching, multimodal sensor fusion, and 3D reconstruction, its robustness and accuracy in complex field environments have been significantly improved. This paper systematically reviews recent progress in agricultural applications of binocular stereo vision, covering system architectures, traditional and deep learning-based stereo matching methods, point cloud reconstruction techniques, and emerging supervision strategies such as 3D Gaussian splatting. It further summarizes key applications, including high-throughput phenotyping, fruit localization and robotic harvesting, weed detection and precision spraying, autonomous navigation, and livestock body condition assessment, highlighting its role in multi-task agricultural perception systems. Finally, the paper discusses major challenges, including low-texture matching difficulty, occlusions in complex environments, cross-domain generalization, real-time lightweight deployment, and limited dataset availability. Future directions are outlined in foundation model-based visual perception, self- and weakly supervised learning, multimodal fusion, and edge-efficient model design, aiming to support large-scale deployment in smart agriculture.\n\n## 1. Introduction\n\nPlant phenotypes refer to the set of observable traits resulting from the interaction between genotype and environment. Accurate acquisition of crop phenotypic information is fundamental for understanding genotype–phenotype–environment interactions and serves as an essential basis for modern crop breeding, precision crop management, and intelligent agricultural decision-making [[1](https:\u002F\u002Fwww.mdpi.com\u002F2076-3417\u002F16\u002F16\u002F7957#B1-applsci-16-07957),[2](https:\u002F\u002Fwww.mdpi.com\u002F2076-3417\u002F16\u002F16\u002F7957#B2-applsci-16-07957),[3](https:\u002F\u002Fwww.mdpi.com\u002F2076-3417\u002F16\u002F16\u002F7957#B3-applsci-16-07957)]. Traditional phenotyping methods rely heavily on manual measurements, such as plant height estimation with rulers, leaf counting, and destructive biomass sampling [[4](https:\u002F\u002Fwww.mdpi.com\u002F2076-3417\u002F16\u002F16\u002F7957#B4-applsci-16-07957),[5](https:\u002F\u002Fwww.mdpi.com\u002F2076-3417\u002F16\u002F16\u002F7957#B5-applsci-16-07957)]. As shown in the study by Zou et al., manual counting in mature wheat density estimation is not only inefficient, but also incurs substantially increasing labor costs as cultivation scale expands [[6](https:\u002F\u002Fwww.mdpi.com\u002F2076-3417\u002F16\u002F16\u002F7957#B6-applsci-16-07957)]. In addition, conventional approaches typically depend on discrete sampling at isolated time points, making it difficult to capture continuous growth dynamics throughout the crop life cycle, which often leads to missing information at key developmental stages.\n\nTwo-dimensional (2D) imaging technologies have partially alleviated the limitations of manual measurements. Unmanned aerial vehicles (UAVs) and satellite remote sensing platforms enable large-scale and high-throughput data acquisition [[7](https:\u002F\u002Fwww.mdpi.com\u002F2076-3417\u002F16\u002F16\u002F7957#B7-applsci-16-07957),[8](https:\u002F\u002Fwww.mdpi.com\u002F2076-3417\u002F16\u002F16\u002F7957#B8-applsci-16-07957)]. However, due to the inherent loss of depth information in 2D projections, the full three-dimensional structure of crops cannot be faithfully represented. Issues such as leaf overlap, organ occlusion, and viewpoint distortion remain difficult to resolve [[9](https:\u002F\u002Fwww.mdpi.com\u002F2076-3417\u002F16\u002F16\u002F7957#B9-applsci-16-07957),[10](https:\u002F\u002Fwww.mdpi.com\u002F2076-3417\u002F16\u002F16\u002F7957#B10-applsci-16-07957),[11](https:\u002F\u002Fwww.mdpi.com\u002F2076-3417\u002F16\u002F16\u002F7957#B11-applsci-16-07957)]. For crops with complex canopy structures, such as maize, wheat, and rice, 2D images are insufficient to accurately characterize key structural traits, including leaf area index, leaf angle distribution, and canopy volume, thereby limiting their applicability in fine-grained phenotypic analysis [[12](https:\u002F\u002Fwww.mdpi.com\u002F2076-3417\u002F16\u002F16\u002F7957#B12-applsci-16-07957),[13](https:\u002F\u002Fwww.mdpi.com\u002F2076-3417\u002F16\u002F16\u002F7957#B13-applsci-16-07957)].\n\nIn parallel with advances in sensing hardware, intelligent analysis methods have become increasingly important for converting agricultural images and three-dimensional data into biologically and operationally meaningful information. Conventional machine-learning methods rely on hand-crafted features and statistical classifiers, whereas recent deep learning approaches can automatically learn hierarchical representations for image classification, object detection, semantic and instance segmentation, stereo matching, and pose estimation. For three-dimensional agricultural data, point-cloud filtering, registration, segmentation, skeleton extraction, geometric modeling, and trait-estimation algorithms are commonly used to characterize plant and animal structures. Multimodal sensor fusion and temporal modeling further integrate complementary information from RGB images, depth sensors, LiDAR, multispectral cameras, and sequential observations. Together, these intelligent analysis methods support crop phenotyping, fruit localization, weed recognition, autonomous navigation, robotic harvesting, and livestock monitoring, thereby transforming raw sensor measurements into actionable information for smart agriculture.\n\nFrom a global perspective, binocular stereo vision and intelligent analysis methods have attracted increasing attention in agricultural research and automation. Studies in North America, Europe, and East Asia have progressively moved from controlled laboratory experiments toward field-based deployment and integrated perception systems for high-throughput phenotyping, robotic harvesting, precision spraying, autonomous machinery, and livestock monitoring. Regional research priorities vary according to crop types, production systems, labor availability, equipment infrastructure, and investment in agricultural digitalization. Meanwhile, the availability of low-cost stereo cameras, deep learning models, edge-computing platforms, and multimodal sensing systems has lowered the technical barriers to agricultural 3D perception. Nevertheless, global adoption remains uneven because of limited domain-specific datasets, substantial environmental variability, differences in farm scale and infrastructure, and the absence of standardized evaluation and deployment frameworks. These conditions indicate that future progress requires not only algorithmic innovation, but also internationally diverse datasets, cross-region validation, affordable hardware integration, and closer collaboration among researchers, equipment manufacturers, and agricultural practitioners.\n\nThe emergence of three-dimensional (3D) perception technologies has provided a promising pathway to overcome these limitations. Point clouds and mesh-based representations preserve complete spatial geometric information, enabling organ-level morphometric measurements and canopy-scale structural analysis [[14](https:\u002F\u002Fwww.mdpi.com\u002F2076-3417\u002F16\u002F16\u002F7957#B14-applsci-16-07957),[15](https:\u002F\u002Fwww.mdpi.com\u002F2076-3417\u002F16\u002F16\u002F7957#B15-applsci-16-07957)]. Wu et al. constructed maize plant point clouds using multi-view stereo vision and achieved the automatic extraction of stem–leaf segmentation, plant height, and leaf area, with a coefficient of determination (R 2) of 0.994 between estimated and manually measured plant height [[16](https:\u002F\u002Fwww.mdpi.com\u002F2076-3417\u002F16\u002F16\u002F7957#B16-applsci-16-07957)]. Zhou et al. further developed a temporal point cloud dataset covering four crops, including maize, soybean, and rice, across eleven varieties [[17](https:\u002F\u002Fwww.mdpi.com\u002F2076-3417\u002F16\u002F16\u002F7957#B17-applsci-16-07957)]. By combining skeleton extraction and leaf matching, they enabled cross-temporal organ tracking and revealed dynamic growth patterns across different crops. Compared with 2D methods, 3D data provide richer structural information and support non-destructive, continuous monitoring, significantly improving measurement consistency and repeatability [[18](https:\u002F\u002Fwww.mdpi.com\u002F2076-3417\u002F16\u002F16\u002F7957#B18-applsci-16-07957),[19](https:\u002F\u002Fwww.mdpi.com\u002F2076-3417\u002F16\u002F16\u002F7957#B19-applsci-16-07957)].\n\nAmong existing 3D sensing technologies, binocular stereo vision occupies a unique position. Current crop 3D sensing approaches can be broadly categorized into active and passive methods [[20](https:\u002F\u002Fwww.mdpi.com\u002F2076-3417\u002F16\u002F16\u002F7957#B20-applsci-16-07957),[21](https:\u002F\u002Fwww.mdpi.com\u002F2076-3417\u002F16\u002F16\u002F7957#B21-applsci-16-07957),[22](https:\u002F\u002Fwww.mdpi.com\u002F2076-3417\u002F16\u002F16\u002F7957#B22-applsci-16-07957)]. Active methods, including structured light, time-of-flight (Time of Flight) cameras, and LiDAR, obtain depth information by actively emitting signals [[23](https:\u002F\u002Fwww.mdpi.com\u002F2076-3417\u002F16\u002F16\u002F7957#B23-applsci-16-07957),[24](https:\u002F\u002Fwww.mdpi.com\u002F2076-3417\u002F16\u002F16\u002F7957#B24-applsci-16-07957)]. Although these methods offer high accuracy, they suffer from high hardware cost and limited adaptability to outdoor environments. In field-based crop phenotyping, ToF cameras are prone to generating large invalid measurement regions under strong ambient illumination, while LiDAR, despite its high precision, remains expensive. Passive reconstruction methods mainly include multi-view stereo (Multi-View Stereo), structure from motion (Structure from Motion), and binocular stereo vision, all of which rely solely on ambient light for 3D reconstruction [[25](https:\u002F\u002Fwww.mdpi.com\u002F2076-3417\u002F16\u002F16\u002F7957#B25-applsci-16-07957),[26](https:\u002F\u002Fwww.mdpi.com\u002F2076-3417\u002F16\u002F16\u002F7957#B26-applsci-16-07957),[27](https:\u002F\u002Fwww.mdpi.com\u002F2076-3417\u002F16\u002F16\u002F7957#B27-applsci-16-07957)].\n\nBinocular stereo vision has become one of the most promising solutions for agricultural 3D perception due to several advantages [[28](https:\u002F\u002Fwww.mdpi.com\u002F2076-3417\u002F16\u002F16\u002F7957#B28-applsci-16-07957),[29](https:\u002F\u002Fwww.mdpi.com\u002F2076-3417\u002F16\u002F16\u002F7957#B29-applsci-16-07957)]. First, it is highly cost-effective, requiring only two standard RGB cameras, with hardware costs far lower than LiDAR systems [[30](https:\u002F\u002Fwww.mdpi.com\u002F2076-3417\u002F16\u002F16\u002F7957#B30-applsci-16-07957)]. Zhang et al. compared fruit tree canopy volume estimation between binocular vision and LiDAR in UAV-based precision spraying and reported that the binocular system reduced cost by more than 80% while maintaining volume estimation errors within 10% [[31](https:\u002F\u002Fwww.mdpi.com\u002F2076-3417\u002F16\u002F16\u002F7957#B31-applsci-16-07957)]. Second, binocular stereo vision provides a practical and cost-effective solution for open-field applications, but its performance remains sensitive to strong and spatially variable illumination. Direct sunlight, shadows, overexposure, and specular reflections may reduce photometric consistency between stereo image pairs and consequently degrade matching accuracy. In comparison, LiDAR is generally more robust to variations in ambient illumination, although intense solar radiation may still reduce its signal-to-noise ratio and effective sensing range. Third, it provides high information density by generating dense depth maps, which are essential for fine-grained organ-level phenotyping. Yang et al. combined binocular stereo vision with supervision signals derived from 3D Gaussian splatting to generate dense depth maps for leafy vegetables, achieving an endpoint error (EPE) of only 0.91 pixels in leaf regions [[32](https:\u002F\u002Fwww.mdpi.com\u002F2076-3417\u002F16\u002F16\u002F7957#B32-applsci-16-07957)]. Fourth, the method benefits from a mature algorithmic foundation, and recent advances in deep learning have further significantly improved stereo matching accuracy [[33](https:\u002F\u002Fwww.mdpi.com\u002F2076-3417\u002F16\u002F16\u002F7957#B33-applsci-16-07957)].\n\nHowever, binocular stereo vision also faces several critical limitations in agricultural applications. The first is the difficulty of stereo matching in low-texture and repetitive-texture regions, as crop leaf surfaces often lack distinctive features while repetitive structures easily lead to mismatches. Wu et al. observed in oil-tea fruit harvesting robots that traditional stereo matching methods produce large holes in leaf regions, whereas improved deep learning-based networks significantly reduce 3D localization errors [[16](https:\u002F\u002Fwww.mdpi.com\u002F2076-3417\u002F16\u002F16\u002F7957#B16-applsci-16-07957)]. Occlusion remains another major challenge. Zou et al. reported that severe occlusion in dense wheat canopies leads to substantial segmentation errors when directly applying clustering methods on reconstructed point clouds [[6](https:\u002F\u002Fwww.mdpi.com\u002F2076-3417\u002F16\u002F16\u002F7957#B6-applsci-16-07957)]. The trade-off between computational complexity and real-time performance is also prominent. Yi et al., in a study on real-time volume estimation of four fruit types, found that lightweight models achieved higher frame rates but underperformed in occlusion handling and small-object detection compared to more complex networks [[34](https:\u002F\u002Fwww.mdpi.com\u002F2076-3417\u002F16\u002F16\u002F7957#B34-applsci-16-07957)]. In addition, high-quality stereo matching datasets for agricultural scenes are extremely limited. Unlike autonomous driving, which benefits from large-scale datasets such as KITTI, existing agricultural datasets are small and lack diversity. To address this issue, Yang et al. constructed a lettuce stereo dataset with 36 scenes and more than 8000 stereo image pairs by leveraging 3D Gaussian splatting for multi-view reconstruction and dense annotation generation [[32](https:\u002F\u002Fwww.mdpi.com\u002F2076-3417\u002F16\u002F16\u002F7957#B32-applsci-16-07957)].\n\nTo address these challenges, recent studies have made substantial progress from the perspectives of algorithm design, dataset construction, and system integration. Niu et al. proposed a stereo foundation model fine-tuning strategy driven by 2D Gaussian splatting for maize field scenarios [[35](https:\u002F\u002Fwww.mdpi.com\u002F2076-3417\u002F16\u002F16\u002F7957#B35-applsci-16-07957)]. Using multi-view reconstruction, they synthesized 3625 high-fidelity stereo pairs, and through a partial parameter freezing strategy, fine-tuned only 45.7% of trainable parameters, reducing end-point error in disparity estimation to 1.85 pixels. Wang et al. introduced the Plant Stereo dataset, providing a high-quality benchmark for plant 3D reconstruction [[36](https:\u002F\u002Fwww.mdpi.com\u002F2076-3417\u002F16\u002F16\u002F7957#B36-applsci-16-07957)]. Ajagekar et al. integrated vision foundation models such as Segment Anything with RGB-D cameras to build a phenotyping system for densely planted lettuce, enabling non-destructive and continuous monitoring of plant height, area, and volume at 10-min intervals, with correlation coefficients exceeding 0.94 compared to manual measurements [[37](https:\u002F\u002Fwww.mdpi.com\u002F2076-3417\u002F16\u002F16\u002F7957#B37-applsci-16-07957)].\n\nThis review provides a comprehensive narrative overview of advances in binocular stereo vision for crop phenotyping and smart agriculture, focusing on representative studies published between 2015 and 2025. This time window was selected because deep learning achieved a major breakthrough in stereo matching around 2015. A structured literature search was conducted using databases including Web of Science, Scopus, and ScienceDirect, with keywords covering the intersection of binocular stereo vision, stereo matching, plant phenotyping, crops, and precision agriculture. Relevant studies were identified through the iterative screening of titles, abstracts, and full texts, with priority given to methodologically innovative works and those reporting quantitative performance evaluations. While we aimed for broad coverage across application domains, we acknowledge that the review is narrative in nature and does not follow formal systematic review protocols (e.g., PRISMA). Given the interdisciplinary nature of the field and the diversity of agricultural scenarios, research designs, and evaluation metrics, we consider this narrative approach most suitable for synthesizing current progress and identifying future research priorities.\n\nOverall, binocular stereo vision has evolved from early proof-of-concept studies toward large-scale application-oriented research in agriculture, exhibiting trends toward higher accuracy, real-time processing, lightweight deployment, and improved generalization. However, a significant gap remains between algorithmic performance and large-scale field deployment, particularly in terms of robustness in complex environments, cross-domain generalization, and the lack of standardized evaluation frameworks. Therefore, a systematic review of existing studies is essential for further advancing this field.\n\n## 2. Fundamentals of Stereo Vision and Adaptation to Agricultural Environments\n\nStereo vision reconstructs three-dimensional spatial structures from two-dimensional images by simulating the human binocular vision mechanism [[38](https:\u002F\u002Fwww.mdpi.com\u002F2076-3417\u002F16\u002F16\u002F7957#B38-applsci-16-07957),[39](https:\u002F\u002Fwww.mdpi.com\u002F2076-3417\u002F16\u002F16\u002F7957#B39-applsci-16-07957)]. It serves as a key technical approach for object localization and point cloud reconstruction in agricultural scenes. Despite its strong potential in agricultural phenotyping, its practical deployment in complex field environments remains challenging due to high model complexity, limited dataset availability, and the absence of unified evaluation standards. This section systematically analyzes the core components of binocular stereo vision and their suitability for agricultural applications from a system-level perspective [[40](https:\u002F\u002Fwww.mdpi.com\u002F2076-3417\u002F16\u002F16\u002F7957#B40-applsci-16-07957),[41](https:\u002F\u002Fwww.mdpi.com\u002F2076-3417\u002F16\u002F16\u002F7957#B41-applsci-16-07957)].\n\n### 2.1. Construction and Computational Pipeline of Binocular Stereo Vision Systems\n\nThe construction and computational pipeline of binocular stereo vision systems consist of three tightly coupled stages: camera calibration, epipolar rectification, and outdoor system design [[42](https:\u002F\u002Fwww.mdpi.com\u002F2076-3417\u002F16\u002F16\u002F7957#B42-applsci-16-07957),[43](https:\u002F\u002Fwww.mdpi.com\u002F2076-3417\u002F16\u002F16\u002F7957#B43-applsci-16-07957)]. These stages jointly determine the final accuracy of depth estimation.\n\nCamera calibration is the foundational step in stereo vision systems, aiming to establish the mapping relationship between pixel coordinates and 3D world coordinates by estimating intrinsic and extrinsic camera parameters [[44](https:\u002F\u002Fwww.mdpi.com\u002F2076-3417\u002F16\u002F16\u002F7957#B44-applsci-16-07957),[45](https:\u002F\u002Fwww.mdpi.com\u002F2076-3417\u002F16\u002F16\u002F7957#B45-applsci-16-07957),[46](https:\u002F\u002Fwww.mdpi.com\u002F2076-3417\u002F16\u002F16\u002F7957#B46-applsci-16-07957)]. The checkerboard-based calibration method proposed by Zhang has been widely adopted in agricultural applications due to its high accuracy and robust implementation [[31](https:\u002F\u002Fwww.mdpi.com\u002F2076-3417\u002F16\u002F16\u002F7957#B31-applsci-16-07957)]. Zou et al. a","MDPI Applied Sciences 2026-08-10","2026-08-10T00:00:00Z","论文",10,false,75,{"impact":17,"substance":18,"depth":17,"authority":19,"freshness":20,"relevant":21,"comment":22},18,20,12,7,1,"系统综述双目立体视觉在农业中的应用，涵盖方法、应用与挑战，对智慧农业有参考价值。",[24],{"name":10,"url":6},[26,27,28,29],"农业机器人","深度学习","立体视觉","3D感知",{"code":31,"slug":32,"name":33,"path":34,"parent_code":35},"I2","agricultural-ai","农业AI与大模型","\u002Ftopic\u002Fdigital-agriculture\u002Fagricultural-ai","I",[],{"status":38,"locked":14,"version":39,"method":40,"evidence":41},"classified","2026-10-22.2","rules",{"candidates":42,"agricultural_evidence":45},[43,49,56],{"code":31,"score":44,"supported":45,"title":46,"abstract":47,"tags":48},4,true,[],[27],[27],{"code":50,"score":51,"supported":45,"title":52,"abstract":53,"tags":55},"C",3,[],[54],"畜禽",[],{"code":57,"score":21,"supported":14,"title":58,"abstract":59,"tags":60},"F6",[],[],[26],[62,63],"农业机器人 深度学习 立体视觉 3D感知","农业机器人 深度学习","农业机器人深度学习立体视觉3D感知-467",5,"2026-08-12T23:57:24.301767Z",{"total":68,"page":21,"page_size":68,"items":69},6,[70,144,201,278,359,440],{"id":71,"title":72,"url":73,"summary":74,"summary_zh":75,"content":8,"source_name":76,"source_url":73,"published_at":77,"category":12,"cover_url":8,"hotness":13,"is_selected":14,"score":78,"score_detail":79,"sources":83,"tags":85,"primary_topic":90,"secondary_topics":96,"classification":98,"search_phrases":120,"slug":123,"view_count":124,"doi":125,"paper":126,"created_at":143},5387,"LIGHTWEIGHT FEATURE FUSED ATTENTION DRIVEN MULTI LEVEL HYBRID MOBILENET V2 FRAMEWORK FOR EARLY TOMATO LEAF DISEASE DETECTION","https:\u002F\u002Fdoi.org\u002F10.62441\u002Factainnovations.vi.1918","It is crucial to find and treat tomato crop diseases early and accurately to improve crop efficiency and make sure there is no scarcity of crop. However, computerized analysis is provocative due to lighting differences, compound backgrounds, and graphic likenesses among infection regions. Research work proposed a lightweight Attention Driven Multi Level Hybrid MobileNet-V2 algorithm that fuses the handcrafted texture structures with Deep Learning (DL) features, then image preprocessing is incorporated using Gaussian, Bilateral, and Guided filters to minimize noise by preserving the disease lesions. LAB color model along with Contrast Limited Adaptive Histogram Equalization (CLAHE) are used to improve the contrast. Otsu thresholding is involved to effectively separate the leaf segment from diseased parts. Local Binary Pattern (LBP) and Gray Level Co-occurrence Matrix (GLCM) descriptors are used to obtain fine-grained features, while high level features are gained by pretrained MobileNet-V2 model. Features fusion is fed to the fully connected layers of deep learning model for training and validation purpose. Results have revealed that the suggested model attains better classification performance as compared to existing models, making it an effective explanation for real time tomato leaf disease detection for smart agriculture applications.","早期准确发现并治疗番茄作物病害，对于提高作物效率、确保作物不短缺至关重要。然而，由于光照差异、复杂背景以及感染区域之间的图形相似性，计算机化分析具有挑战性。研究提出了一种轻量级注意力驱动多层混合MobileNet-V2算法，该算法将手工纹理结构与深度学习（DL）特征相融合，然后结合高斯、双边和引导滤波器进行图像预处理，以在保留病害病变的同时最小化噪声。采用LAB颜色模型和对比度受限自适应直方图均衡化（CLAHE）来提高对比度。引入Otsu阈值分割以有效分离叶片片段与病害部分。使用局部二值模式（LBP）和灰度共生矩阵（GLCM）描述符获取细粒度特征，同时通过预训练的MobileNet-V2模型获取高层特征。特征融合后输入深度学习模型的全连接层进行训练和验证。结果表明，与现有模型相比，所提出的模型取得了更好的分类性能，使其成为智慧农业应用中实时番茄叶片病害检测的有效解决方案。","Acta Innovations","2026-10-18T00:00:00Z",64,{"impact":19,"substance":17,"depth":80,"authority":13,"freshness":81,"relevant":21,"comment":82},16,8,"提出融合手工纹理特征与MobileNet-V2的轻量番茄叶病检测方法，有一定方法新颖性，但属单篇论文、应用范围有限，未达每日精选门槛。",[84],{"name":76,"url":73},[86,87,27,88,89],"智慧农业","农业人工智能","番茄病害","图像识别",{"code":91,"slug":92,"name":93,"path":94,"parent_code":95},"B1","vegetables","蔬菜","\u002Ftopic\u002Fhorticulture\u002Fvegetables","B",[97],{"code":31,"slug":32,"name":33,"path":34,"parent_code":35},{"status":38,"locked":14,"version":39,"method":40,"evidence":99},{"candidates":100,"agricultural_evidence":45},[101,108,114],{"code":91,"score":102,"supported":45,"title":103,"abstract":105,"tags":107},15,[104],"tomato",[106,104],"番茄",[88],{"code":31,"score":109,"supported":45,"title":110,"abstract":111,"tags":113},9,[],[27,112],"deep learning",[87,89,27],{"code":115,"score":20,"supported":45,"title":116,"abstract":117,"tags":119},"I1",[],[86,118],"smart agriculture",[86],[121,122],"MobileNet-V2 番茄叶片病害","番茄叶病 深度学习 检测","MobileNet-V2番茄叶片病害-5387",0,"10.62441\u002Factainnovations.vi.1918",{"doi":125,"openalex_id":127,"authors":128,"venue":76,"cited_by_count":124,"oa_url":134,"card":135,"direction":141,"ingested_from":142},"W7222977034",[129,131],{"name":130,"orcid":8},"Mrs. Vijetalaxmi A Hegde",{"name":132,"orcid":133},"Abhay Ramakant Kasetwar","https:\u002F\u002Forcid.org\u002F0000-0002-5800-8764","https:\u002F\u002Factainnovations.com\u002Findex.php\u002Fpub\u002Farticle\u002Fdownload\u002F1918\u002F981",{"tldr":136,"method":137,"finding":138,"direction":139,"opportunity":140},"提出轻量注意力多级混合MobileNet-V2框架，融合手工纹理与深度学习特征实现番茄叶片病害早期检","高斯\u002F双边\u002F引导滤波去噪，LAB+CLAHE增强，Otsu分割，LBP与GLCM","所提模型分类性能优于现有模型，可有效用于智慧农业实时番茄叶片病害检测。","农业人工智能与决策模型","可探索多作物跨域泛化、田间复杂光照下轻量模型部署与可解释性，及病害早期分级检测。","智慧农业 \u002F 农业物联网","openalex","2026-10-20T23:30:10.424511Z",{"id":145,"title":146,"url":147,"summary":148,"summary_zh":149,"content":8,"source_name":150,"source_url":147,"published_at":77,"category":12,"cover_url":8,"hotness":13,"is_selected":14,"score":151,"score_detail":152,"sources":154,"tags":156,"primary_topic":159,"secondary_topics":160,"classification":161,"search_phrases":186,"slug":189,"view_count":124,"doi":190,"paper":191,"created_at":200},5385,"A Comprehensive Review of Deep Learning and Computer Vision Techniques for Fruit Quality Assessment and Classification","https:\u002F\u002Fdoi.org\u002F10.25258\u002Fijddt.16.75s.151","Fruit image analysis has become an essential research area in precision agriculture, enabling automated fruit classification, quality grading, maturity estimation, disease diagnosis, yield prediction, and intelligent harvesting through advances in computer vision and artificial intelligence (AI).Recent progress in imaging technologies, large-scale benchmark datasets, and deep learning has substantially improved the accuracy, robustness, and scalability of automated fruit analysis systems.This survey presents a comprehensive and systematic review of image processing and AI techniques for fruit image analysis.It covers the complete image analysis pipeline, including image acquisition, preprocessing, enhancement, color space transformation, segmentation, feature extraction, and handcrafted texture descriptors.Furthermore, conventional machine learning algorithms are reviewed alongside state-of-the-art deep learning architectures, including Convolutional Neural Networks (CNNs), Vision Transformers (ViTs), and Vision Mamba models, for applications such as fruit recognition, quality assessment, disease detection, ripeness estimation, and object detection.Publicly available benchmark datasets, evaluation metrics, and performance comparison strategies are also summarized to facilitate objective assessment of existing approaches.In addition, the survey discusses key research challenges, including illumination variations, occlusion, complex orchard environments, limited annotated datasets, computational complexity, and domain adaptation, while highlighting emerging research trends such as self-supervised learning, multimodal learning, explainable AI, lightweight deep networks, foundation models, and edge intelligence for real-time deployment.By providing a unified review of methodologies, datasets, evaluation protocols, challenges, and future research directions, this survey offers a comprehensive reference for researchers and practitioners developing intelligent and scalable fruit image analysis systems for precision agriculture.","水果图像分析已成为精准农业中的一个重要研究领域，借助计算机视觉和人工智能（AI）的进步，实现了果实自动分类、品质分级、成熟度估计、病害诊断、产量预测和智能采收。成像技术、大规模基准数据集和深度学习的最新进展显著提升了自动化水果分析系统的准确性、鲁棒性和可扩展性。本综述对用于水果图像分析的图像处理和AI技术进行了全面而系统的回顾。内容涵盖完整的图像分析流程，包括图像采集、预处理、增强、颜色空间变换、分割、特征提取和手工纹理描述符。此外，综述了传统机器学习算法以及最先进的深度学习架构，包括卷积神经网络（CNN）、视觉Transformer（ViT）和视觉Mamba模型，用于果实识别、品质评估、病害检测、成熟度估计和目标检测等应用。同时总结了公开可用的基准数据集、评价指标和性能比较策略，以便对现有方法进行客观评估。此外，本综述讨论了关键研究挑战，包括光照变化、遮挡、复杂果园环境、标注数据集有限、计算复杂度和域适应，并重点介绍了新兴研究趋势，如自监督学习、多模态学习、可解释AI、轻量级深度网络、基础模型和面向实时部署的边缘智能。通过提供方法、数据集、评价协议、挑战和未来研究方向的统一综述，本综述为开发面向精准农业的智能且可扩展的水果图像分析系统的研究人员和实践者提供了全面的参考。","International Journal of Drug Delivery Technology",56,{"impact":81,"substance":17,"depth":80,"authority":68,"freshness":81,"relevant":21,"comment":153},"系统性综述覆盖CNN、ViT、Vision Mamba等前沿方法，信息量足，但期刊权威性一般且偏学术综述，公共价值有限，适合主题聚合而非每日精选头条。",[155],{"name":150,"url":147},[86,87,157,27,158],"计算机视觉","水果品质检测",{"code":31,"slug":32,"name":33,"path":34,"parent_code":35},[],{"status":38,"locked":14,"version":39,"method":40,"evidence":162},{"candidates":163,"agricultural_evidence":45},[164,174,180],{"code":31,"score":165,"supported":45,"title":166,"abstract":168,"tags":173},28,[112,167],"computer vision",[169,27,170,157,171,112,172,167],"目标检测","机器学习","machine learning","artificial intelligence",[87,27,157],{"code":115,"score":20,"supported":45,"title":175,"abstract":176,"tags":179},[],[177,178],"精准农业","precision agriculture",[86],{"code":181,"score":51,"supported":45,"title":182,"abstract":183,"tags":185},"B2",[],[184],"orchard",[],[187,188],"水果品质 计算机视觉 深度学习","农业人工智能 水果品质检测 计算机视觉 智慧农业","水果品质计算机视觉深度学习-5385","10.25258\u002Fijddt.16.75s.151",{"doi":190,"openalex_id":192,"authors":193,"venue":150,"cited_by_count":124,"oa_url":147,"card":194,"direction":141,"ingested_from":142},"W7222892831",[],{"tldr":195,"method":196,"finding":197,"direction":198,"opportunity":199},"系统综述深度学习与计算机视觉在水果品质评估与分级中的全流程方法。","综述图像采集到分割、特征提取及CNN、ViT、Vision Mamba等模型与公","深度学习显著提升水果分析精度与鲁棒性，但光照、遮挡和小样本仍是主要瓶颈。","农业遥感与作物表型","可探索自监督与轻量化模型在复杂果园遮挡、小样本场景下的边缘实时部署。","2026-10-20T23:30:10.066042Z",{"id":202,"title":203,"url":204,"summary":205,"summary_zh":206,"content":8,"source_name":207,"source_url":204,"published_at":208,"category":12,"cover_url":8,"hotness":13,"is_selected":14,"score":209,"score_detail":210,"sources":212,"tags":214,"primary_topic":218,"secondary_topics":223,"classification":225,"search_phrases":250,"slug":253,"view_count":21,"doi":254,"paper":255,"created_at":277},5165,"Proximal Remote Sensing of Oil Palm Basal Stem Rot Using Deep Learning Vision Algorithms and Ground-based Images","https:\u002F\u002Fdoi.org\u002F10.22266\u002Fijies2026.1130.31","Basal Stem Rot (BSR), caused by Ganoderma boninense, threatens oil palm productivity in Southeast Asia.This study develops and benchmarks a low-cost remote sensing framework for BSR severity diagnosis from smartphone-captured RGB trunk images, evaluating two pathways for future field deployment.The first pairs YOLO detection with a Random Forest classifier: YOLO26n gave the best accuracy-to-cost trade-off among three nanovariants for localizing BSR symptoms (mushroom, rot, white-button), and the end-to-end YOLO26n+RF pipeline achieved 0.83 accuracy and macro-F1 of 0.79 across Healthy, Mild, Moderate, and Severe classes.Rot was consistently the hardest symptom to localize, owing to its diffuse, boundary-ambiguous appearance.The second pathway trained four image classifiers (EfficientNet-B0, MobileNetV2, Swin-Tiny, and ConvNeXt-V2-Tiny) directly on severity labels, with ConvNeXt-V2-Tiny performing best (0.83 accuracy, 0.83 macro-F1).Both pathways performed well at severity extremes but struggled with Moderate, reflecting visual overlap between stages.SHAP analysis identified rot-area ratio and fruiting-body count as top predictors.These offline results show accurate BSR diagnosis is potentially achievable from smartphone imagery, laying groundwork for future on-device deployment.","由灵芝（Ganoderma boninense）引起的基茎腐病（Basal Stem Rot, BSR）威胁着东南亚油棕的生产力。本研究开发并基准测试了一个低成本遥感框架，用于从智能手机拍摄的RGB树干图像中诊断基茎腐病的严重程度，并评估了未来田间部署的两条技术路径。第一条路径将YOLO检测与随机森林分类器相结合：在三种纳米变体中，YOLO26n在定位基茎腐病症状（蘑菇、腐烂、白钮）方面实现了最佳的精度-成本权衡，端到端的YOLO26n+RF流程在健康、轻度、中度和重度四个类别上达到了0.83的准确率和0.79的宏F1值。腐烂始终是最难定位的症状，原因在于其弥散且边界模糊的外观。第二条路径直接在严重程度标签上训练了四个图像分类器（EfficientNet-B0、MobileNetV2、Swin-Tiny和ConvNeXt-V2-Tiny），其中ConvNeXt-V2-Tiny表现最佳（准确率0.83，宏F1值0.83）。两条路径在严重程度极端类别上表现良好，但在中度类别上表现欠佳，反映出各阶段之间存在视觉重叠。SHAP分析确定腐烂面积比和子实体数量为最重要的预测因子。这些离线结果表明，利用智能手机图像有望实现准确的基茎腐病诊断，为未来设备端部署奠定了基础。","International journal of intelligent engineering and systems","2026-10-16T00:00:00Z",69,{"impact":19,"substance":18,"depth":80,"authority":19,"freshness":109,"relevant":21,"comment":211},"该研究利用智能手机图像与深度学习实现油棕基腐病严重度诊断，方法新颖、结论可靠，但属细分作物研究，产业影响有限，适合作为技术类精选。",[213],{"name":207,"url":204},[86,215,27,216,217],"农业遥感","油棕","病害诊断",{"code":219,"slug":220,"name":221,"path":222,"parent_code":35},"I3","remote-sensing","遥感与农情监测","\u002Ftopic\u002Fdigital-agriculture\u002Fremote-sensing",[224],{"code":31,"slug":32,"name":33,"path":34,"parent_code":35},{"status":38,"locked":14,"version":39,"method":40,"evidence":226},{"candidates":227,"agricultural_evidence":45},[228,234,239,246],{"code":219,"score":102,"supported":45,"title":229,"abstract":231,"tags":233},[230],"remote sensing",[232,230],"遥感",[215],{"code":31,"score":19,"supported":45,"title":235,"abstract":236,"tags":238},[112],[237],"YOLO",[27],{"code":240,"score":68,"supported":45,"title":241,"abstract":242,"tags":245},"B5",[],[243,244],"蘑菇","mushroom",[],{"code":115,"score":21,"supported":14,"title":247,"abstract":248,"tags":249},[],[],[86],[251,252],"油棕 基腐病 遥感","Ganoderma boninense 智能手机 检测","油棕基腐病遥感-5165","10.22266\u002Fijies2026.1130.31",{"doi":254,"openalex_id":256,"authors":257,"venue":207,"cited_by_count":124,"oa_url":204,"card":272,"direction":198,"ingested_from":142},"W7222626135",[258,261,264,266,269],{"name":259,"orcid":260},"Wan Siu Hong","https:\u002F\u002Forcid.org\u002F0000-0002-8254-4962",{"name":262,"orcid":263},"Jonathan Yong Chung Ee","https:\u002F\u002Forcid.org\u002F0000-0003-4079-8005",{"name":265,"orcid":8},"Jin Yuan Chan",{"name":267,"orcid":268},"Mohamad Izzuddin Anuar","https:\u002F\u002Forcid.org\u002F0000-0002-4929-5175",{"name":270,"orcid":271},"Mahmud Iwan Solihin","https:\u002F\u002Forcid.org\u002F0000-0002-5293-7466",{"tldr":273,"method":274,"finding":275,"direction":198,"opportunity":276},"用手机拍摄油棕树干图像，结合深度学习检测与分类，实现基腐病严重度诊断。","YOLO26n+随机森林与四种图像分类器，基于手机RGB图像，SHAP解释。","两条路径准确率均约0.83，中度病级最难区分，腐区比例与子实体数为关键预测因子。","可探索轻量化模型在手机端离线部署，并扩充多地区、多品种数据以提升中度病级辨识。","2026-10-17T23:30:28.307953Z",{"id":279,"title":280,"url":281,"summary":282,"summary_zh":283,"content":8,"source_name":207,"source_url":281,"published_at":208,"category":12,"cover_url":8,"hotness":13,"is_selected":14,"score":209,"score_detail":284,"sources":286,"tags":288,"primary_topic":291,"secondary_topics":297,"classification":299,"search_phrases":320,"slug":323,"view_count":21,"doi":324,"paper":325,"created_at":358},5164,"Lightweight Hybrid CNN-Transformer Network for Robust Maize Leaf Disease Detection and Edge Deployment","https:\u002F\u002Fdoi.org\u002F10.22266\u002Fijies2026.1130.50","Early and accurate detection of maize leaf diseases is essential for improving crop productivity and supporting precision agriculture.However, deep learning models trained on controlled leaf-image datasets often experience performance degradation under real-world agricultural conditions because of illumination variation, background clutter, occlusion, leaf orientation, and imaging noise.To address these challenges, this study develops a lightweight deep learning framework for robust maize leaf disease classification and edge-oriented deployment.The framework employs MobileNetV3-Small for efficient extraction of disease-discriminative visual features and investigates lightweight feature-learning strategies for improving classification under challenging field conditions.Experiments were conducted using PlantVillage-Maize and Field-Maize datasets comprising 6,188 images from four categories: Maize Leaf Blight (MLB), Common Rust, Healthy, and Leaf Spot.The image-only MobileNetV3-Small model achieved 97.8% accuracy and a 97.4% F1-score on the controlled PlantVillage-Maize dataset, while its performance decreased to 86.1% accuracy and an 83.9% F1-score on Field-Maize, highlighting the domain gap between controlled and real-field imagery.Robustness was further evaluated under Gaussian noise, illumination variation, and partial occlusion, together with computational-efficiency and edge-deployment analyses.The findings demonstrate the importance of lightweight and robust visual feature learning for practical maize disease diagnosis under real-world agricultural imaging conditions.","早期准确检测玉米叶片病害对于提高作物生产力和支持精准农业至关重要。然而，在受控叶片图像数据集上训练的深度学习模型，在真实农业条件下往往会出现性能下降，原因包括光照变化、背景杂乱、遮挡、叶片朝向以及成像噪声。为应对这些挑战，本研究开发了一个轻量级深度学习框架，用于稳健的玉米叶片病害分类和面向边缘的部署。该框架采用MobileNetV3-Small高效提取具有病害判别力的视觉特征，并研究轻量级特征学习策略，以提升在复杂田间条件下的分类性能。实验使用PlantVillage-Maize和Field-Maize数据集进行，共包含来自四个类别的6，188张图像：玉米大斑病（Maize Leaf Blight，MLB）、普通锈病（Common Rust）、健康（Healthy）和叶斑病（Leaf Spot）。仅使用图像的MobileNetV3-Small模型在受控的PlantVillage-Maize数据集上达到了97.8%的准确率和97.4%的F1分数，而在Field-Maize上的性能下降至86.1%的准确率和83.9%的F1分数，凸显了受控图像与真实田间图像之间的域差距。研究进一步在高斯噪声、光照变化和部分遮挡条件下评估了稳健性，并进行了计算效率和边缘部署分析。研究结果表明，轻量级且稳健的视觉特征学习对于真实农业成像条件下的实际玉米病害诊断具有重要意义。",{"impact":19,"substance":18,"depth":80,"authority":19,"freshness":109,"relevant":21,"comment":285},"该研究提出轻量级玉米叶病检测框架，并揭示受控数据集与田间数据之间的性能差距，方法扎实、结论有参考价值，但属细分技术论文，公共影响有限。",[287],{"name":207,"url":281},[86,87,289,27,290],"边缘计算","玉米病害",{"code":292,"slug":293,"name":294,"path":295,"parent_code":296},"A3","maize","玉米","\u002Ftopic\u002Ffield-crops\u002Fmaize","A",[298],{"code":31,"slug":32,"name":33,"path":34,"parent_code":35},{"status":38,"locked":14,"version":39,"method":40,"evidence":300},{"candidates":301,"agricultural_evidence":45},[302,307,311,315],{"code":292,"score":303,"supported":45,"title":304,"abstract":305,"tags":306},14,[293],[294,293],[],{"code":31,"score":81,"supported":45,"title":308,"abstract":309,"tags":310},[],[27,112],[87,27],{"code":115,"score":20,"supported":45,"title":312,"abstract":313,"tags":314},[],[177,178],[86],{"code":316,"score":21,"supported":14,"title":317,"abstract":318,"tags":319},"I5",[],[],[289],[321,322],"MobileNetV3 玉米叶部病害","玉米叶片病害 边缘部署","MobileNetV3玉米叶部病害-5164","10.22266\u002Fijies2026.1130.50",{"doi":324,"openalex_id":326,"authors":327,"venue":207,"cited_by_count":124,"oa_url":281,"card":353,"direction":198,"ingested_from":142},"W7222602850",[328,331,334,336,338,341,343,345,348,350],{"name":329,"orcid":330},"Awwab Mohammad","https:\u002F\u002Forcid.org\u002F0000-0002-1148-5678",{"name":332,"orcid":333},"Punith kumar","https:\u002F\u002Forcid.org\u002F0009-0005-5676-4194",{"name":335,"orcid":8},"Bhavani K",{"name":337,"orcid":8},"Sailee N. Panchbhai",{"name":339,"orcid":340},"Sagarkumar S. Badhiye","https:\u002F\u002Forcid.org\u002F0000-0002-0710-3761",{"name":342,"orcid":8},"Manoj Kumar S B",{"name":344,"orcid":8},"Dharamvir",{"name":346,"orcid":347},"Sumit Kumar Sar","https:\u002F\u002Forcid.org\u002F0000-0003-0384-9088",{"name":349,"orcid":8},"Rani Chhagan Dalvi",{"name":351,"orcid":352},"Yogesh Harishchandra Bhosale","https:\u002F\u002Forcid.org\u002F0000-0001-6901-1419",{"tldr":354,"method":355,"finding":356,"direction":139,"opportunity":357},"构建轻量MobileNetV3-Small框架实现玉米叶片病害分类并评估边缘部署。","MobileNetV3-Small，PlantVillage-Maize与Fie","受控数据准确率97.8%，田间降至86.1%，揭示明显域差距。","可研究跨域自适应与田间鲁棒特征学习，缩小受控与真实场景性能差距。","2026-10-17T23:30:28.248539Z",{"id":360,"title":361,"url":362,"summary":363,"summary_zh":364,"content":8,"source_name":365,"source_url":362,"published_at":208,"category":12,"cover_url":8,"hotness":13,"is_selected":14,"score":366,"score_detail":367,"sources":372,"tags":374,"primary_topic":377,"secondary_topics":382,"classification":384,"search_phrases":400,"slug":403,"view_count":21,"doi":404,"paper":405,"created_at":439},5150,"MSFANet: a multi-scale feature aggregation network for tobacco pest recognition","https:\u002F\u002Fdoi.org\u002F10.3389\u002Ffpls.2026.1871587","Introduction Tobacco is a globally significant economic crop, and pest infestation represents one of the primary factors contributing to declines in both yield and quality. However, tobacco pests encompass a wide variety of species that differ substantially in morphology and scale. Field-captured images are frequently subject to complex background interference and class imbalance, presenting considerable challenges for automated recognition. Existing deep learning approaches remain limited in their capacity for multi-scale feature extraction and discriminative feature representation under such conditions. To address these challenges, this study proposes a Multi-Scale Feature Aggregation Network (MSFANet) for tobacco pest recognition, formulated here as a single-label image classification task. Methods To address scale variation and complex field backgrounds, two task-oriented modules were developed. The Multi-Scale Pyramid Pooling (MSPP) module aggregates contextual information across multiple receptive fields through parallel feature extraction and adaptive cross-branch fusion. The Hierarchical Feature Fusion Block (HFFB) subsequently refines the feature representations through spatial gating and partial convolution. An auxiliary classification head and Focal Loss were incorporated during training to provide intermediate supervision and address class imbalance, respectively. Results The effectiveness of MSFANet was evaluated on a dataset comprising 13 tobacco-associated pest and natural-enemy categories. The model achieved accuracy, precision, recall, and F1-score of 97.63%, 97.89%, 97.59%, and 97.71%, respectively, with a parameter count of 1.21 M. Among the models trained from random initialization, MSFANet shared the highest accuracy while using the fewest parameters. Discussion By combining multi-scale contextual aggregation with hierarchical feature refinement, MSFANet achieved accurate classification across the 13 tobacco-associated categories under complex field conditions. The proposed method provides effective technical support for automated and precision-oriented pest monitoring in tobacco fields, offering practical value for advancing precise pest management and the broader development of smart agriculture.","引言 烟草是全球重要的经济作物，虫害侵染是导致其产量和品质下降的主要因素之一。然而，烟草害虫种类繁多，形态和尺度差异显著。田间采集的图像常受到复杂背景干扰和类别不平衡的影响，给自动化识别带来了相当大的挑战。现有深度学习方法在此类条件下进行多尺度特征提取和判别性特征表示的能力仍然有限。为应对这些挑战，本研究提出了一种用于烟草害虫识别的多尺度特征聚合网络（Multi-Scale Feature Aggregation Network，MSFANet），并将其形式化为单标签图像分类任务。方法 针对尺度变化和复杂田间背景，开发了两个面向任务的模块。多尺度金字塔池化（Multi-Scale Pyramid Pooling，MSPP）模块通过并行特征提取和自适应跨分支融合，聚合多个感受野的上下文信息。分层特征融合模块（Hierarchical Feature Fusion Block，HFFB）随后通过空间门控和部分卷积对特征表示进行细化。训练过程中引入辅助分类头和Focal Loss，分别提供中间监督和处理类别不平衡。结果 在包含13个烟草相关害虫及天敌类别的数据集上评估了MSFANet的有效性。模型的准确率、精确率、召回率和F1分数分别为97.63%、97.89%、97.59%和97.71%，参数量为1.21 M。在随机初始化训练的模型中，MSFANet以最少的参数量取得了最高准确率。讨论 通过结合多尺度上下文聚合与分层特征细化，MSFANet在复杂田间条件下实现了对13个烟草相关类别的准确分类。所提方法为烟草田间害虫的自动化和精准监测提供了有效的技术支持，对推进精准害虫管理和智慧农业的更广泛发展具有实用价值。","Frontiers in Plant Science",76,{"impact":80,"substance":368,"depth":369,"authority":370,"freshness":109,"relevant":21,"comment":371},21,17,13,"该论文提出轻量化多尺度特征聚合网络，在烟草害虫识别上取得97%以上精度且参数量仅1.21M，方法新颖、数据扎实，对智慧农业虫情监测有实用价值，值得进入每日精选。",[373],{"name":365,"url":362},[86,87,27,375,376],"病虫害识别","烟草",{"code":378,"slug":379,"name":380,"path":381,"parent_code":296},"A6","cash-crops","棉麻糖与特色经作","\u002Ftopic\u002Ffield-crops\u002Fcash-crops",[383],{"code":31,"slug":32,"name":33,"path":34,"parent_code":35},{"status":38,"locked":14,"version":39,"method":40,"evidence":385},{"candidates":386,"agricultural_evidence":45},[387,392,396],{"code":378,"score":102,"supported":45,"title":388,"abstract":390,"tags":391},[389],"tobacco",[376,389],[376],{"code":31,"score":109,"supported":45,"title":393,"abstract":394,"tags":395},[],[27,112],[87,27,375],{"code":115,"score":20,"supported":45,"title":397,"abstract":398,"tags":399},[],[86,118],[86],[401,402],"MSFANet 烟草害虫识别","多尺度特征聚合 害虫识别","MSFANet烟草害虫识别-5150","10.3389\u002Ffpls.2026.1871587",{"doi":404,"openalex_id":406,"authors":407,"venue":365,"cited_by_count":124,"oa_url":433,"card":434,"direction":141,"ingested_from":142},"W7222603139",[408,410,412,414,416,418,420,422,425,428,431],{"name":409,"orcid":8},"Shun He",{"name":411,"orcid":8},"Sifan Li",{"name":413,"orcid":8},"Xinjian Li",{"name":415,"orcid":8},"Qianqian Luo",{"name":417,"orcid":8},"Dong Sufen",{"name":419,"orcid":8},"Lu Lei",{"name":421,"orcid":8},"Xingjiang Chen",{"name":423,"orcid":424},"Liuti Cai","https:\u002F\u002Forcid.org\u002F0000-0002-9151-8002",{"name":426,"orcid":427},"Jian‐Yu Meng","https:\u002F\u002Forcid.org\u002F0000-0002-5796-1086",{"name":429,"orcid":430},"Ning Lu","https:\u002F\u002Forcid.org\u002F0000-0003-0214-5038",{"name":432,"orcid":8},"Hancheng Wang","https:\u002F\u002Fwww.frontiersin.org\u002Fjournals\u002Fplant-science\u002Farticles\u002F10.3389\u002Ffpls.2026.1871587\u002Fpdf",{"tldr":435,"method":436,"finding":437,"direction":139,"opportunity":438},"提出MSFANet多尺度特征聚合网络，实现复杂田间背景下烟草害虫的自动识别。","多尺度金字塔池化MSPP、分层特征融合HFFB、辅助分类头与Focal Loss","模型准确率97.63%、参数量仅1.21M，在随机初始化模型中精度最高且参数最少。","可探索轻量多尺度网络在边缘设备部署及跨作物害虫迁移识别，结合类不平衡与复杂背景的鲁棒性研究。","2026-10-17T23:30:08.695455Z",{"id":441,"title":442,"url":443,"summary":444,"summary_zh":8,"content":8,"source_name":445,"source_url":8,"published_at":446,"category":12,"cover_url":8,"hotness":13,"is_selected":14,"score":15,"score_detail":447,"sources":449,"tags":451,"primary_topic":454,"secondary_topics":455,"classification":456,"search_phrases":467,"slug":470,"view_count":21,"doi":471,"paper":472,"created_at":480},5062,"Detection of yam leaf diseases and pest damage under complex field conditions using LIMN-YOLO","https:\u002F\u002Fwww.frontiersin.org\u002Fjournals\u002Fplant-science\u002Farticles\u002F10.3389\u002Ffpls.2026.1955275\u002Ffull","Bibo Lu、Yuanqi Ren(河南理工大学)、Jie Yang(焦作农林科学研究院特色农业研究所)、Yanjun Lu(北京航空航天大学\u002F河南理工大学)在《Frontiers in Plant Science》2026-10-12接受的论文,基于YOLO11n提出LIMN-YOLO轻量级山药叶部病害及虫害检测模型,在田间遮挡与不均匀光照等复杂条件下的小目标识别难题进行针对性改进。模型引入非均匀退化增强、响应引导跨尺度融合与尺度敏感的边界框回归,在包含1,434张图像、9,221个标注实例的四类山药病害数据集上,mAP@0.5与mAP@0.5:0.95分别达到0.8792与0.7182,较基线YOLO11n显著提升;三次随机种子对比实验显示模型稳定性较好。","Frontiers in Plant Science 16 2026-10-12","2026-10-12T00:00:00Z",{"impact":102,"substance":368,"depth":17,"authority":370,"freshness":81,"relevant":21,"comment":448},"基于YOLO11n的轻量级山药叶部病害检测模型，方法有针对性改进且实验数据扎实，属细分领域技术进展。",[450],{"name":445,"url":443},[86,87,27,452,453],"山药病害检测","小目标检测",{"code":31,"slug":32,"name":33,"path":34,"parent_code":35},[],{"status":38,"locked":14,"version":39,"method":40,"evidence":457},{"candidates":458,"agricultural_evidence":45},[459,463],{"code":31,"score":370,"supported":45,"title":460,"abstract":461,"tags":462},[237],[237],[87,27],{"code":115,"score":21,"supported":14,"title":464,"abstract":465,"tags":466},[],[],[86],[468,469],"河南理工大学 山药 病害检测","LIMN-YOLO 山药 叶片病害","河南理工大学山药病害检测-5062","10.3389\u002Ffpls.2026.1955275\u002Ffull",{"doi":471,"openalex_id":8,"authors":473,"venue":8,"cited_by_count":124,"oa_url":8,"card":474,"direction":139,"ingested_from":479},[],{"tldr":475,"method":476,"finding":477,"direction":139,"opportunity":478},"提出轻量级LIMN-YOLO模型，在复杂田间条件下检测山药叶部病害与虫害。","基于YOLO11n改进，引入非均匀退化增强、跨尺度融合与尺度敏感回归，用1434","mAP@0.5达0.8792，mAP@0.5:0.95达0.7182，较基线显著提升且稳定性好。","可探索轻量模型在移动端\u002F边缘设备的实时部署，并扩展至更多作物与多病害联合检测。","agent","2026-10-16T00:14:01.055092Z"]