Smart agricultural technology最新文献_第9页

Integrating deep learning and mobile imaging for assessment of automated conformational indices and weight prediction in Brahman cattle 整合深度学习和移动成像，评估婆罗门牛的自动构象指数和体重预测

IF 6.3

Smart agricultural technology Pub Date : 2025-06-06 DOI: 10.1016/j.atech.2025.101079

Peerayut Nilchuen , Thanathip Suwanasopee , Skorn Koonawootrittriron

{"title":"Integrating deep learning and mobile imaging for assessment of automated conformational indices and weight prediction in Brahman cattle","authors":"Peerayut Nilchuen , Thanathip Suwanasopee , Skorn Koonawootrittriron","doi":"10.1016/j.atech.2025.101079","DOIUrl":"10.1016/j.atech.2025.101079","url":null,"abstract":"<div><div>Accurate, non-invasive assessment of cattle body conformation and weight is critical for advancing productivity and genetic improvement in tropical beef production systems. Conventional methods are labor-intensive, require large equipment, involve direct animal contact, cause stress to animals, and are often impractical for smallholders under resource-limited conditions and lacking proper infrastructure. This study presents a novel, smartphone-based system for real-time body measurement and weight estimation in Brahman cattle using a cloud-integrated artificial intelligence (AI) model. A total of 12,660 side-view images were collected and annotated for hip depth (HD) and body length (BL), with YOLOv11 convolutional neural network variants trained and validated. The YOLOv11m model demonstrated the best performance (precision: 99.85%, recall: 100%, F1-score: 99.92%, IoU: 97.68 ± 1.31%), with automated measurements showing strong agreement with manual ImageJ data (MAPE < 0.4%). HD and BL were highly correlated in both sexes (<em>r</em> = 0.98–0.99) and moderately predictive of body weight (<em>r</em> = 0.57–0.59). A multiple regression model using HD and BL achieved the highest prediction accuracy for body weight (MAE = 43.44 kg; MAPE = 8.91%). The system was deployed through a LINE messaging chatbot app, enabling users to submit cattle images and receive instant measurements and weight estimations directly via smartphone–eliminating the need for specialized hardware. This low-cost, user-friendly AI tool offers a scalable solution for digital phenotyping, livestock monitoring, and informed selection in smallholder settings. The approach holds strong potential to support data-driven decision-making and sustainable productivity gains in tropical beef production systems.</div></div>","PeriodicalId":74813,"journal":{"name":"Smart agricultural technology","volume":"12 ","pages":"Article 101079"},"PeriodicalIF":6.3,"publicationDate":"2025-06-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144280781","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Two–stage multimodal 3D point localization framework for automatic grape harvesting 葡萄自动收获的两阶段多模态三维点定位框架

IF 6.3

Smart agricultural technology Pub Date : 2025-06-06 DOI: 10.1016/j.atech.2025.101062

Qian Shen , Dayu Xu , Tianyu Guo , Xiaobo Mao , Fang Xia

{"title":"Two–stage multimodal 3D point localization framework for automatic grape harvesting","authors":"Qian Shen , Dayu Xu , Tianyu Guo , Xiaobo Mao , Fang Xia","doi":"10.1016/j.atech.2025.101062","DOIUrl":"10.1016/j.atech.2025.101062","url":null,"abstract":"<div><div>This study proposes a lightweight Two–Stage multimodal 3D point localization framework for automated grape harvesting, addressing the challenge of precise 3D harvesting point localization. Unlike traditional methods, it employs a Two–Stage multimodal fusion framework, linking RGB and depth images. In the first–stage, pedicels in RGB images are segmented to generate masks. To tackle missing depth information and outliers, an Adaptive Percentile Filtering and Irregular Group-Based Completion (APF–IGBC) algorithm is proposed, leveraging depth distribution patterns and morphological features of grape pedicels. Guided by the mask, APF–IGBC efficiently filters and complements depth information. In the second stage, semantic features from the mask are integrated into the depth image via the Inward Shrinkage Method (ISM) for pose estimation, extracting three key points on pedicels for precise 3D localization. The framework enhances depth restoration and pose estimation accuracy through multimodal fusion. To address multi-scale pedicel challenges, Shared Self–learning YOLO (SSL–YOLO) is introduced, utilizing a Shared Self–learning Head (SSL–Head) for cross-scale information flow. SSL–YOLO achieves 103.9 FPS (9.8 GFLOPs, 2.7M Params) in instance segmentation and 118.8 FPS (6.1 GFLOPs, 2.6M Params) in pose estimation, demonstrating lightweight efficiency, with AP@50 scores of 99.1% and 99.5%, respectively. In comprehensive experiments on a self-constructed grape dataset, the framework achieves a P of 99.2% and a R of 99.2% for 3D harvesting point localization within 600 mm. It has a computational cost of 15.9 GFLOPs and 5.3M Params, running at 100.6 FPS on a GPU and 27.6 FPS on a CPU, showcasing high accuracy and practicality.</div></div>","PeriodicalId":74813,"journal":{"name":"Smart agricultural technology","volume":"12 ","pages":"Article 101062"},"PeriodicalIF":6.3,"publicationDate":"2025-06-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144262444","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

EinsteinNet and state-of-the-art ML models for android-based orange classification: Integration, evaluation, and deployment 用于基于android的橙色分类的einstein和最先进的ML模型：集成、评估和部署

IF 6.3

Smart agricultural technology Pub Date : 2025-06-06 DOI: 10.1016/j.atech.2025.101072

Ashif Ahmed Shuvo, Wahada Jinnat Oishy Bhuian, Afzal Rahman, Abdullah Iqbal

{"title":"EinsteinNet and state-of-the-art ML models for android-based orange classification: Integration, evaluation, and deployment","authors":"Ashif Ahmed Shuvo, Wahada Jinnat Oishy Bhuian, Afzal Rahman, Abdullah Iqbal","doi":"10.1016/j.atech.2025.101072","DOIUrl":"10.1016/j.atech.2025.101072","url":null,"abstract":"<div><div>The integration of on-device machine learning (ML) into mobile platforms has the potential to enable intelligent, real-time diagnostics in agricultural settings. This study presents EinsteinNet, a lightweight convolutional neural network (CNN) optimized for offline orange quality classification on Android devices. A custom dataset of 15,000 annotated images across five quality categories—fresh, rotten, green, canker-affected, and black-spotted—was used to train and compare EinsteinNet against four established architectures (ResNet50, DenseNet121, MobileNetV2, NASNetMobile) and a no-code Google Teachable Machine baseline. EinsteinNet achieved 99.6 % test accuracy with a compact model size (254 KB), but incurred higher inference latency (∼1118 ms) relative to other models. All networks were converted to TensorFlow Lite (TFLite) format and integrated into an Android application with full offline inference capabilities. Empirical evaluation on a Google Pixel 6 showed that while custom CNNs offer strong classification performance and deployment efficiency, optimizing for real-time responsiveness remains critical. Power consumption metrics collected via Android Profiler revealed critical trade-offs among inference accuracy, latency, and energy usage, underscoring the balance required in deploying edge AI models for precision agriculture.</div></div>","PeriodicalId":74813,"journal":{"name":"Smart agricultural technology","volume":"12 ","pages":"Article 101072"},"PeriodicalIF":6.3,"publicationDate":"2025-06-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144262445","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Blockchain-enabled traceability and certification for frozen food supply chains: A conceptual design 冷冻食品供应链的区块链可追溯性和认证：概念设计

IF 6.3

Smart agricultural technology Pub Date : 2025-06-06 DOI: 10.1016/j.atech.2025.101085

Havva Uyar , Athanasios Papanikolaou , Evgenia Kapassa , Marios Touloupos , Stamatia Rizou

{"title":"Blockchain-enabled traceability and certification for frozen food supply chains: A conceptual design","authors":"Havva Uyar , Athanasios Papanikolaou , Evgenia Kapassa , Marios Touloupos , Stamatia Rizou","doi":"10.1016/j.atech.2025.101085","DOIUrl":"10.1016/j.atech.2025.101085","url":null,"abstract":"<div><div>Ensuring traceability, compliance certification and cold chain integrity in frozen food supply chains remains a persistent challenge, exacerbated by fragmented monitoring systems, manual audits and vulnerability to data manipulation. This study presents a conceptual design for a blockchain-enabled compliance architecture that addresses these challenges by integrating real-time Internet of Things (IoT) data acquisition, permissioned blockchain-based data storage and smart contract-driven compliance automation. Following a Design Science Research (DSR) methodology, the research focuses on the initial phases (problem identification, objective specification and artefact conceptualization) providing a structured foundation for future demonstration and evaluation. The proposed design is structured across three interdependent layers: (1) a Data Acquisition Layer that ensures continuous and secure sensor-based monitoring; (2) a Data Storage Layer that leverages blockchain for immutable recording and transparent auditability; and (3) an Application Layer that integrates smart contracts for automated compliance enforcement and user interfaces for stakeholder interaction. By translating regulatory compliance requirements into a modular, blockchain-based design, this work contributes to the theoretical grounding of decentralized regulatory infrastructures in agri-food systems. The proposed architecture embodies design principles that may inform similar traceability systems across other regulated supply chains. Although empirical validation is forthcoming, the conceptualization serves as a scaffold for future DSR iterations and contributes to design knowledge in the domain of digital compliance architectures.</div></div>","PeriodicalId":74813,"journal":{"name":"Smart agricultural technology","volume":"12 ","pages":"Article 101085"},"PeriodicalIF":6.3,"publicationDate":"2025-06-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144271567","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Adaptive path planning for efficient object search by UAVs in agricultural fields 农业无人机高效目标搜索的自适应路径规划

IF 6.3

Smart agricultural technology Pub Date : 2025-06-06 DOI: 10.1016/j.atech.2025.101075

Rick van Essen , Eldert van Henten , Lammert Kooistra , Gert Kootstra

{"title":"Adaptive path planning for efficient object search by UAVs in agricultural fields","authors":"Rick van Essen , Eldert van Henten , Lammert Kooistra , Gert Kootstra","doi":"10.1016/j.atech.2025.101075","DOIUrl":"10.1016/j.atech.2025.101075","url":null,"abstract":"<div><div>This paper presents an adaptive path planner for object search in agricultural fields using UAVs. The path planner uses a high-altitude coverage flight path and plans additional low-altitude inspections when the detection network is uncertain. The path planner was evaluated in an offline simulation environment containing real-world images. We trained a YOLOv8 detection network to detect artificial plants placed in grass fields to showcase the potential of our path planner. We evaluated the effect of different detection certainty measures, optimized the path planning parameters, investigated the effects of localization errors, and different numbers of objects in the field. The YOLOv8 detection confidence worked best to differentiate between true and false positive detections and was therefore used in the adaptive planner. The optimal parameters of the path planner depended on the distribution of objects in the field. When the objects were uniformly distributed, more low-altitude inspections were needed compared to a non-uniform distribution of objects, resulting in a longer path length. The adaptive planner proved to be robust against localization uncertainty. When increasing the number of objects, the flight path length increased, especially when the objects were uniformly distributed. When the objects were non-uniformly distributed, the adaptive path planner yielded a shorter path than a low-altitude coverage path, even with a high number of objects. Overall, the presented adaptive path planner allowed finding non-uniformly distributed objects in a field faster than a coverage path planner and resulted in a compatible detection accuracy. The path planner is made available at <span><span>https://github.com/wur-abe/uav_adaptive_planner</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":74813,"journal":{"name":"Smart agricultural technology","volume":"12 ","pages":"Article 101075"},"PeriodicalIF":6.3,"publicationDate":"2025-06-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144242499","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Algorithms in the orchard: An embedding-based expert answering system for apple rust 果园中的算法：基于嵌入式的苹果锈病专家应答系统

IF 6.3

Smart agricultural technology Pub Date : 2025-06-06 DOI: 10.1016/j.atech.2025.101069

Astha Anand , Jian Shen , Armin Bernd Cremers , Marc Jacobs

引用次数: 0

A near real-time spatial decision support system for improving sugarcane monitoring through a satellite mapping web browser 通过卫星地图网络浏览器改进甘蔗监测的近实时空间决策支持系统

IF 6.3

Smart agricultural technology Pub Date : 2025-06-06 DOI: 10.1016/j.atech.2025.101084

Bryan Alemán-Montes , Pere Serra , Alaitz Zabala , Joan Masó , Xavier Pons

{"title":"A near real-time spatial decision support system for improving sugarcane monitoring through a satellite mapping web browser","authors":"Bryan Alemán-Montes , Pere Serra , Alaitz Zabala , Joan Masó , Xavier Pons","doi":"10.1016/j.atech.2025.101084","DOIUrl":"10.1016/j.atech.2025.101084","url":null,"abstract":"<div><div>The global importance of sustainable sugarcane as a source of food and energy has driven the development of decision-making tools based on remote sensing (RS) to improve crop management. An approach in agricultural lands is the implementation of spatial decision support systems (S-DSS) for crop monitoring. However, most of these systems are designed for global or regional scales, limiting their applicability to local contexts with specific requirements. This study proposes a methodology to address some weaknesses associated with the underuse of S-DSS by integrating end-user requirements into the design process. To achieve this an easy-to-use near real-time S-DSS was developed, tailored to the needs of two sugarcane cooperatives in Costa Rica, validated with real data and field work, and adapted to three management scales (cooperative, farm and plot). Our Sugarcane Satellite Tracking (SugarSaT) provides two core tools: sugarcane harvest progress monitoring and an early warning system. The results validated that SugarSaT offers a suitable approach for the monitoring of sugarcane plantations that uses current and historical satellite data. Regarding the harvested area, more than 93 % of plots was correctly identified when 100 % of the sugarcane was delivered to the mill whereas about the early warning system, a plot test considering anomalies caused by bloom achieved an overall accuracy of 75.3 %. A usability test revealed that 83 % of the surveyed agronomic advisors believed that SugarSaT is suitable for integration into their daily activities. In conclusion, this S-DSS represents a significant step forward in sugarcane monitoring, enabling agronomic advisors to leverage satellite imagery for spatially informed decision-making while balancing scientific insights with the practical needs of end-users.</div></div>","PeriodicalId":74813,"journal":{"name":"Smart agricultural technology","volume":"12 ","pages":"Article 101084"},"PeriodicalIF":6.3,"publicationDate":"2025-06-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144253479","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Design optimization and aerodynamic investigations of air suction seed metering systems through CFD-DEM approach 基于CFD-DEM的空气吸收式排种系统设计优化及气动特性研究

IF 6.3

Smart agricultural technology Pub Date : 2025-06-06 DOI: 10.1016/j.atech.2025.101082

Saddam Hussain , Yong Chen , Xing Yu , Muhammad Usman Farid , Abdul Ghafoor , Salah Jumaa Alshamali , Taj Munir , Jianjun Hu

{"title":"Design optimization and aerodynamic investigations of air suction seed metering systems through CFD-DEM approach","authors":"Saddam Hussain , Yong Chen , Xing Yu , Muhammad Usman Farid , Abdul Ghafoor , Salah Jumaa Alshamali , Taj Munir , Jianjun Hu","doi":"10.1016/j.atech.2025.101082","DOIUrl":"10.1016/j.atech.2025.101082","url":null,"abstract":"<div><div>The current study focuses on the design optimization of air-suction seed metering devices for precision agriculture. The effect of vacuum pressure, suction hole diameter, and seed disk speed on the performance of a metering system for corn precision seeder was investigated using the Computational Fluid Dynamics (CFD) approach. The key parameters were modeled and optimized using both single-factor analysis and response surface methodology. The results highlighted the critical role of suction holes in generating rapid pressure drops, facilitating efficient seed pickup and adhesion. The velocity and pressure contours indicated that well-optimized settings ensure stable suction, smooth airflow, and accurate seed handling. The optimal parameter combination comprising vacuum pressure of 3 kPa, suction hole diameter of 4 mm, and seed disk rotation speed of 30 RPM achieved the maximum pressure difference and improved system stability. This combination was further validated using CFD-DEM coupling for a single seed. The analysis revealed that the proposed design not only minimizes seed-to-seed interference but also improves precise seeding. The study optimized the air-suction precision seeder by conducting a single-factor analysis to determine the optimal ranges for vacuum pressure, operating speed, and suction hole diameter. The orthogonal factor testing further refined the parameters, resulting in a 3.5 kPa vacuum pressure, 7 km/h operating speed, and a 4 mm suction hole diameter as the optimal combination. The bench test results confirmed the accuracy of the optimization with M<sub>1</sub> = 95.98%, M<sub>2</sub> = 1.5%, and M<sub>3</sub> = 2.52%. This research provides a foundation and strong justification for improving air-suction seed metering systems, thereby significantly enhancing precision seeder efficiency and crop productivity.</div></div>","PeriodicalId":74813,"journal":{"name":"Smart agricultural technology","volume":"12 ","pages":"Article 101082"},"PeriodicalIF":6.3,"publicationDate":"2025-06-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144271566","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Machine learning techniques for tomato yield prediction: A comprehensive analysis 番茄产量预测的机器学习技术：综合分析

IF 6.3

Smart agricultural technology Pub Date : 2025-06-06 DOI: 10.1016/j.atech.2025.101067

Kodjo Abel Odah , Sèton Calmette Ariane Houetohossou , Vinasetan Ratheil Houndji , Romain Lucas Glèlè Kakaï

{"title":"Machine learning techniques for tomato yield prediction: A comprehensive analysis","authors":"Kodjo Abel Odah , Sèton Calmette Ariane Houetohossou , Vinasetan Ratheil Houndji , Romain Lucas Glèlè Kakaï","doi":"10.1016/j.atech.2025.101067","DOIUrl":"10.1016/j.atech.2025.101067","url":null,"abstract":"<div><div>Effective yield prediction is crucial for farmers and the agricultural sector. It allows producers to enhance control over their operations and better align with market supply and demand. With the emergence of Artificial Intelligence (AI), various Machine Learning (ML) models have been developed to predict crop yield. In this study, we conducted a systematic literature review to examine the ML models used for predicting tomato yield, the features associated with the most effective models, and the challenges faced by users. We retrieved 1,486 scientific papers from six electronic databases. Following the PRISMA guidelines, we included 57 studies in our analysis. The results showed that 66.67% of the models achieving the best performance in predicting or estimating tomato yield are Deep Learning (DL) models, with neural networks accounting for 42.11% of these. Specifically, Long Short-Term Memory (LSTM), Artificial Neural Networks (ANN), and Support Vector Regression (SVR) are the models most commonly used, demonstrating strong performance when considering factors such as climate, soil conditions, plant growth, fertilization, and irrigation. Additionally, when using computed vegetation indices from image data, Random Forest Regression (RFR) is frequently applied with notable success. The YOLO-Tomato and R-CNN methods are commonly used for detecting tomato fruits prior to yield estimation. Furthermore, DeepSort and linear regression are the predominant methods employed for counting and estimating tomato yield. For future research, it is important to conduct a comparative analysis of models such as LSTM, ANN, SVR, and RFR specifically for predicting tomato yield using data from Africa.</div></div>","PeriodicalId":74813,"journal":{"name":"Smart agricultural technology","volume":"12 ","pages":"Article 101067"},"PeriodicalIF":6.3,"publicationDate":"2025-06-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144253477","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

EDV-CS-LinkNet: A lightweight semantic segment model of underwater fish school for real-time feeding behaviour quantification in aquaculture EDV-CS-LinkNet：用于水产养殖实时摄食行为量化的水下鱼群轻量级语义段模型

IF 6.3

Smart agricultural technology Pub Date : 2025-06-06 DOI: 10.1016/j.atech.2025.101078

Huihui Yu , Huihui Liu , Zhennan Liu , Zheng Luo , Daoliang Li , Yingyi Chen

{"title":"EDV-CS-LinkNet: A lightweight semantic segment model of underwater fish school for real-time feeding behaviour quantification in aquaculture","authors":"Huihui Yu , Huihui Liu , Zhennan Liu , Zheng Luo , Daoliang Li , Yingyi Chen","doi":"10.1016/j.atech.2025.101078","DOIUrl":"10.1016/j.atech.2025.101078","url":null,"abstract":"<div><div>Quantifying fish school feeding intensity is crucial for intelligent decision-making in feeding strategies. Real-time and precision semantic segmentation of fish and special distribution characteristics of fish school are essential for feeding behaviours quantification. The loss of spatial details and feature of fish school boundary caused by the uneven illumination and free-swimming fish are the main challenges in available deep convolution network models. In this study, an EDV-CS-LinkNet model is proposed for semantic segment model of underwater fish school to quantify the feeding intensity. It improves the LinkNet method by integrating cross-scale features to make a remarkable balance between accuracy and speed. Specifically, the model employs lightweight encoder-decoder variants (EDV) to extract feature maps and introduces cross-stage skip connections (CS) to encode rich spatial features, addressing under- and over-segmentation issues. Additionally, a special feature fusion module (FFM) is introduced to merge shallow and deep image features. Extensive experimental results demonstrate that the proposed method effectively overcomes the challenges of complex underwater environment and free-swimming fish for underwater fish segmentation. The model achieves an accuracy of 95.3 % IOU with an inference speed of 37 FPS. And, it excels in real-time underwater fish segmentation, enabling precise quantification of feeding intensity in intelligent aquaculture.</div></div>","PeriodicalId":74813,"journal":{"name":"Smart agricultural technology","volume":"12 ","pages":"Article 101078"},"PeriodicalIF":6.3,"publicationDate":"2025-06-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144253480","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0