Computers and Electronics in Agriculture最新文献

{"title":"A free-space dielectric system with X-band coaxial-to-waveguide adapters for nondestructive fertility detection in unincubated chicken eggs: Optimizing spectrum, orientation, features, and classifiers","authors":"Niloufar Akbarzadeh ,&nbsp;Seyed Ahmad Mireei ,&nbsp;Gholam Reza Askari ,&nbsp;Mohammad Sedghi ,&nbsp;Abbas Hemmat","doi":"10.1016/j.compag.2025.110447","DOIUrl":"10.1016/j.compag.2025.110447","url":null,"abstract":"<div><div>Detecting infertile eggs before incubation can significantly improve hatch rates and reduce losses associated with billions of infertile eggs. This study employed a free-space dielectric setup with X-band coaxial-to-waveguide adapters to assess egg fertility prior to incubation. Scattering parameters within the 8–12 GHz microwave spectrum were analyzed in both reflectance and transmittance modes, with eggs examined in two distinct orientations. Each sample orientation generated eight spectra, which were preprocessed using various techniques. Seven classifiers were applied to differentiate fertile from infertile eggs, resulting in 576 classification models aimed at identifying the optimal spectrum, sample orientation, preprocessing method, and classifier for distinguishing unincubated fertile and infertile eggs. The insertion loss spectrum in S₂₁ mode (IL_S₂₁) in the vertical orientation was identified as the optimal condition. Several feature selection methods were then evaluated to determine the most informative frequencies. Predictive models were developed using artificial neural networks (ANN), random forest (RF), and boosted trees (BT), leveraging the selected effective frequencies. Notably, the competitive adaptive reweighted sampling (CARS) approach consistently outperformed other methods, yielding robust BT models with an exceptional F1-score of 100 %. In the BT model, CART-based features achieved a sensitivity of 96.00 %, specificity of 93.55 %, precision of 92.31 %, accuracy of 94.64 %, and an F1-score of 94.12 %, comparable to the BT model based on full spectral data (F1-score of 98.04 %). A trade-off exists between the higher accuracy of CARS-selected features and the more localized frequency selection in the CART-based approach. This study highlights the effectiveness of free-space dielectric setups in reliably distinguishing fertile from infertile eggs prior to incubation, offering substantial implications for the poultry industry.</div></div>","PeriodicalId":50627,"journal":{"name":"Computers and Electronics in Agriculture","volume":"236 ","pages":"Article 110447"},"PeriodicalIF":7.7,"publicationDate":"2025-04-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143859624","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"An interpretable temperature prediction method for grain in storage based on improved temporal Fusion Transformers","authors":"Xinze Li ,&nbsp;Wenfu Wu ,&nbsp;Hongpeng Guo ,&nbsp;Xinghan Qiao ,&nbsp;Yanhui Lu ,&nbsp;Yun Wu ,&nbsp;Guoran Qiao","doi":"10.1016/j.compag.2025.110414","DOIUrl":"10.1016/j.compag.2025.110414","url":null,"abstract":"<div><div>Accurate temperature prediction for grain in storage is crucial for safety monitoring and early warning of abnormal conditions. Existing methods for predicting grain temperature face several challenges, such as neglecting spatial dependencies between grain temperatures at different locations within the grain pile, low accuracy, insufficient consideration of multifactorial influences, poor generalization capabilities, and lack of interpretability. To address these challenges, an interpretable temperature prediction model for grain in storage is proposed based on Temporal Fusion Transformers (TFT) integrated with a Graph Convolutional Network (GCN) module. This integration enables the model to simultaneously capture both spatial and temporal dependencies of grain temperatures. The model processes historical grain temperatures, meteorological data, granary internal air temperature and humidity, grain moisture content, grain varieties, and other granary information, categorizing these into static and dynamic variables. The inclusion of weather forecast data for the granary location as known future variables significantly improves prediction accuracy. The interpretability of the model allows for the visualization of input variable importance rankings. Comparative experiments demonstrate that the proposed GCN-TFT model outperforms other comparable models. Practical application experiments further confirm the model’s applicability and effectiveness in predicting grain temperatures. Furthermore, the use of an interpretable model signifies a significant advancement in grain temperature prediction. The interpretable results are expected to assist granary managers in developing effective grain storage management strategies.</div></div>","PeriodicalId":50627,"journal":{"name":"Computers and Electronics in Agriculture","volume":"236 ","pages":"Article 110414"},"PeriodicalIF":7.7,"publicationDate":"2025-04-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143855485","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"An optimization method for corn planting parameters based on mutation breeding sea horse optimization algorithm","authors":"Jinling Bei ,&nbsp;Jiquan Wang ,&nbsp;Hongyu Zhang","doi":"10.1016/j.compag.2025.110417","DOIUrl":"10.1016/j.compag.2025.110417","url":null,"abstract":"<div><div>In response to the black box problem of optimizing corn planting parameters, and considering the shortcomings of traditional methods such as low fitting accuracy and susceptibility to local optima, a MBSHO-BPNN method based on mutation breeding seahorse optimization algorithm (MBSHO) and BP neural network (BPNN) is proposed. Firstly, the standard sea horse optimization algorithm (SHO) and its improved MBSHO are introduced, including control parameter improvement, spiral movement update, and mutation breeding mechanism. Subsequently, a series of extensive numerical experiments are conducted to systematically evaluate the effectiveness of MBSHO components and related parameters. MBSHO is also compared to other algorithms using the CEC 2017 tests on problems of different sizes. The findings indicated that MBSHO demonstrated superior performance. In the effective verification of MBSHO-BPNN, this method outperforms other comparative methods to fit accuracy and optimization results for unconstrained and linearly constrained optimization problems. Ultimately, the MBSHO − BPNN was applied to the optimization of corn planting parameters, and the optimal parameter combination was obtained: the planting density is 9.23 × 10⁴/hm², nitrogen fertilizer application rate is 138.72 kg/hm²,phosphorus fertilizer application rate is 86.53 kg/hm², and the potassium fertilizer application rate is 70.32 kg/hm². Under this configuration, the corn yield reached 16,303.56 kg/hm², which is significantly higher than that of other methods. The relative error of the actual average yield is only − 0.6757 %. This method not only provides an efficient solution to the agricultural black-box optimization problem but also exhibits potential for broader nonlinear optimization challenges.</div></div>","PeriodicalId":50627,"journal":{"name":"Computers and Electronics in Agriculture","volume":"236 ","pages":"Article 110417"},"PeriodicalIF":7.7,"publicationDate":"2025-04-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143855487","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Robust keypoint-based method for peduncle pose estimation in unstructured environments","authors":"Guozhao Shi ,&nbsp;Fugui Zhang ,&nbsp;Xuemei Wu","doi":"10.1016/j.compag.2025.110380","DOIUrl":"10.1016/j.compag.2025.110380","url":null,"abstract":"<div><div>Visual detection for automated fruit harvesting in unstructured environments constitutes a critical technical challenge, especially for fruit peduncles, which exhibit greater sensitivity to environmental factors than the fruits themselves. To address this challenge, this paper proposes a top-down keypoint detection method for pepper peduncles in unstructured environments. The proposed method enables accurate estimation of peduncle poses. The first step of the research involves validating different object detection models and employing ones to identify the bounding boxes of pepper peduncles. Subsequently, a new keypoint detection model based on the Lite Vision Transformer is proposed, leveraging the Transformer’s capacity to capture long-range spatial and semantic dependencies. Experimental results on the pepper dataset collected in unstructured environments demonstrate that the proposed model achieves an AP⁵⁰ of 94.6 %. This performance surpasses multiple state-of-the-art keypoint detection methods while maintaining lightweight parameters and low computational complexity. Moreover, a series of tests reveals that the proposed method outperforms other algorithms in complex environments, especially in occlusion scenarios. Finally, a comprehensive evaluation of the top-down approach is conducted, examining the influence of object detection and keypoint detection models on overall performance. The proposed keypoint detection model achieves the highest performance, with a detection speed of 9.38 FPS when using YOLOv8s as the object detection model, and an AP⁵⁰ of 83.6 % when using YOLOv8l. Experiments conducted in real unstructured environments demonstrated the robustness of the proposed method, effectively detecting the posture of dense and occluded chili pepper peduncles. This research can be extended to the detection of fruit peduncles in other crops, providing a foundation for pose estimation of fruit peduncles in complex environments.</div></div>","PeriodicalId":50627,"journal":{"name":"Computers and Electronics in Agriculture","volume":"236 ","pages":"Article 110380"},"PeriodicalIF":7.7,"publicationDate":"2025-04-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143852120","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Assessment and transferability analysis of soil total nitrogen with different particle sizes based on proximal hyperspectral imaging","authors":"Minyi Zhao ,&nbsp;Zhentao Wang ,&nbsp;Guoqing Chen ,&nbsp;Zhenyang Lv ,&nbsp;Rui Xu ,&nbsp;Yanling Yin ,&nbsp;Jinfeng Wang","doi":"10.1016/j.compag.2025.110409","DOIUrl":"10.1016/j.compag.2025.110409","url":null,"abstract":"<div><div>Hyperspectral imaging serves as a powerful method for conducting efficient and non-invasive detection of total nitrogen in soil. Nevertheless, its complete potential remains underutilized due to the significant need for annotated samples and the influence of variations in soil spectral properties associated with different particle sizes on the generalization capacity of the model. Therefore, this paper proposes a Transfer Component Analysis Adaptive Enhanced Convolutional Neural Network (TACNN), enabling the transfer of soil total nitrogen assessment models between different soil particle size datasets. Six transferability strategies were developed to address diverse application scenarios, and the performance of TACNN was evaluated against TASVR, TAElman, as well as classical ensemble learning models AdaBoost-CNN, AdaBoost-SVR, and AdaBoost-Elman across six soil particle size datasets. The results indicate that classical ensemble learning models achieve satisfactory estimation of soil total nitrogen within the same particle size soil dataset, but fail to transfer across different particle size soil datasets. The combination of TACNN integration with model update demonstrates enhanced capability in estimating soil total nitrogen content across diverse particle size datasets. This research highlights the potential of transfer learning to reduce the dependence of soil total nitrogen assessment models on extensive sample datasets, thereby improving their generalization performance.</div></div>","PeriodicalId":50627,"journal":{"name":"Computers and Electronics in Agriculture","volume":"235 ","pages":"Article 110409"},"PeriodicalIF":7.7,"publicationDate":"2025-04-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143850304","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Orchard variable rate spraying method and experimental study based on multidimensional prescription maps","authors":"Chenming Hu ,&nbsp;Yu Ru ,&nbsp;Shuping Fang ,&nbsp;Zifan Rong ,&nbsp;Hongping Zhou ,&nbsp;Xianghai Yan ,&nbsp;Mengnan Liu","doi":"10.1016/j.compag.2025.110379","DOIUrl":"10.1016/j.compag.2025.110379","url":null,"abstract":"<div><div>The existing prescription map spraying methods only integrate the spraying needs of target crops in a two-dimensional space, neglecting the variations in spraying requirements in depth information, and ignoring the coupling issues between prescription map design and mechanical parameters. This study concentrates on orchard environments and proposes a multi-dimensional prescription map design method based on point cloud data, aimed at guiding variable-rate spraying operations for sprayers. This method includes three main aspects: tree point cloud leaf separation, nozzle position topology and spraying unit division, and multi-dimensional prescription map design. First, the features of the point cloud are enhanced, and the tree leaf point cloud is extracted using a Long Short Term Memory (LSTM) recurrent neural network. Next, the nozzle positions are designed based on canopy structure characteristics, and the canopy area is divided to obtain spraying units. Then, combining the spraying units with wind speed and dosage information, a multi-dimensional prescription map is generated. Finally, the prescription map’s throttle and solenoid valve control strategies are tested and optimized using Hardware In Loop (HIL) methods. The prescription map was applied to orchard spraying, and the experimental results demonstrated that in the deposition monitoring area, the droplet deposition on the front part of the canopy achieved excellent results,. In the rear part of the canopy, 85.7 % of the areas had droplet deposition amounts exceeding 1.2 μL cm⁻². The final experimental results verified the feasibility of the multi-dimensional prescription map, providing theoretical support for the development of intelligent air-assisted spraying equipment for orchards.</div></div>","PeriodicalId":50627,"journal":{"name":"Computers and Electronics in Agriculture","volume":"235 ","pages":"Article 110379"},"PeriodicalIF":7.7,"publicationDate":"2025-04-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143847455","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Adaptive spatial aggregation and viewpoint alignment for three-dimensional online multiple fish tracking","authors":"Yiran Liu ,&nbsp;Beibei Li ,&nbsp;Dingshuo Liu ,&nbsp;Qingling Duan","doi":"10.1016/j.compag.2025.110408","DOIUrl":"10.1016/j.compag.2025.110408","url":null,"abstract":"<div><div>Three-dimensional (3D) multi-object tracking can simultaneously capture the movement trajectories of multiple fish, which is essential for understanding and analysing their movements and behavioural patterns in 3D space. It also provides essential data for applications such as water-quality monitoring, disease diagnosis, and ecological assessment. However, the multi-object tracking of fish in 3D space requires data associations across different perspectives. Variations in scale and appearance across perspectives can lead to inaccurate object positioning and low identification rates. In response to these challenges, in this study, an online 3D multi-object tracking method for fish is proposed based on adaptive spatial aggregation and viewpoint alignment. Dynamic deformable convolution networks (DCNv3) and upsampling techniques are employed to adaptively fuse the fixed-scale features generated by the backbone network, addressing the difficulties in object positioning caused by scale differences. The trajectories of the fish from both the top and side views are then obtained using a cascade tracker. Finally, a viewpoint-alignment approach is proposed to reconstruct the trajectories in 3D space using the two-dimensional (2D) trajectories, thereby avoiding the identity recognition issues caused by drastic changes in appearance. In verifying the effectiveness of the proposed algorithm on the 3D-ZeF20 zebrafish dataset, multi-object tracking accuracy (MOTA) reached 95.03 %; identification F1-score (IDF1) was 97.40 %; and monotonic mean time between failures (MTBFm) was 172 frames. The results demonstrate that this method addresses the difficulties in cross-view matching caused by changes in appearance and scale differences. It enables the simultaneous acquisition of fish multi-object trajectories from front view, top view, and in 3D space, thereby achieving precise online tracking of multiple fish.</div></div>","PeriodicalId":50627,"journal":{"name":"Computers and Electronics in Agriculture","volume":"236 ","pages":"Article 110408"},"PeriodicalIF":7.7,"publicationDate":"2025-04-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143850527","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"TomPhenoNet: A multi-modal fusion and multi-task learning network model for monitoring growth parameters of dwarf tomatoes","authors":"Xunyi Ma ,&nbsp;Yanxu Wu ,&nbsp;Zhixian Lin,&nbsp;Tao Lin","doi":"10.1016/j.compag.2025.110387","DOIUrl":"10.1016/j.compag.2025.110387","url":null,"abstract":"<div><div>Dwarf tomatoes, with high edible and ornamental value, require monitoring multiple growth parameters to balance yield and aesthetics. While deep learning has been widely applied in phenotype monitoring, most studies focus on individual growth parameters, overlooking intrinsic relationships. To simultaneously monitor multiple growth parameters across the entire growth stage and different cultivars, this study develops a multi-modal multi-task phenotype monitoring network for dwarf tomatoes (TomPhenoNet). The network model utilizes top-view RGB-D images to evaluate four key growth parameters: height, leaf area, fresh weight, and the number of red fruits. TomPhenoNet generates mask images, fruit detection features, and the number of detected fruits based on RGB images. By fusing RGB-D images, mask images, and fruit detection features, and introducing the cross-stitch network, the network predicts plant height, leaf area, and fresh weight. The predicted values are further used to generate the dynamic occlusion coefficient, adjusting the number of detected fruits to accurately predict the number of red fruits. Results reveal that TomPhenoNet achieves high prediction performances, with R² values of 0.828, 0.930, 0.945, and 0.881 for plant height, leaf area, fresh weight, and the number of red fruits, respectively. Ablation experiments show that the cross-stitch network and fruit detection features improve the prediction performances of growth parameters, with TomPhenoNet combining both modules performing best. Feature importance analysis indicates the network model captures plant growth characteristics and corrects the impact of leaf occlusion from the top view. This study promotes accurate tomato monitoring and provides data support for optimizing cultivation strategies.</div></div>","PeriodicalId":50627,"journal":{"name":"Computers and Electronics in Agriculture","volume":"235 ","pages":"Article 110387"},"PeriodicalIF":7.7,"publicationDate":"2025-04-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143847322","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"SFCE-VT: Spatial feature fusion and contrast-enhanced visual transformer for fine-grained agricultural pests visual classification","authors":"JianPing Liu ,&nbsp;Lulu Sun ,&nbsp;Guomin Zhou ,&nbsp;Jian Wang ,&nbsp;Jialu Xing ,&nbsp;Chenyang Wang","doi":"10.1016/j.compag.2025.110371","DOIUrl":"10.1016/j.compag.2025.110371","url":null,"abstract":"<div><div>Climate change has led to the intensification of agricultural pests, which are diverse and difficult to identify accurately, and fine-grained classification of agricultural pests is an important method to effectively prevent and control the increasing number of pests, and to ensure the stability and sustainable development of agricultural production. Agricultural pest species can be accurately recognized using deep learning, but current problems such as the small scale agricultural pest data, single scene, and relatively coarse classification results bring challenges to fine-grained image classification of agricultural pests. Therefore, a visual transformer based on spatial feature fusion and contrast enhancement (SFCE-VT) is proposed for fine-grained image classification(FGIC) methods for agricultural pests. First, to accurately localize to the target location, two images, the foreground target, and the occluded background, are cropped using the self-attention mechanism to form three image inputs to complement the detail representation. To further distinguish the foreground target from the background noise, the inputs of three different images are utilized to compare the loss values to enhance the model’s ability to distinguish the foreground target from the background. In addition, to address the challenge of pest recognition from different viewpoints, a self-attention mechanism and graph convolutional network (GCN) are utilized to extract spatial contextual information of the pest region and learn the spatial gesture features of the pests. The experimental results achieved significant performance improvement on both CUB-200-2011 and A-pests, a reconstructed agricultural fine-grained pest dataset, by 1.95% and 3.23% compared to the base vit, respectively. The effectiveness of the cropping contrast enhancement and spatial information learning modules in paying attention to fine-grained features and enriching pest feature information is demonstrated. The source code is publicly available at <span><span>https://github.com/193lulu/SFCE-VT</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":50627,"journal":{"name":"Computers and Electronics in Agriculture","volume":"236 ","pages":"Article 110371"},"PeriodicalIF":7.7,"publicationDate":"2025-04-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143850436","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Machine vision-based assessment of fall color changes in apple leaves and its relationship with nitrogen concentration","authors":"Achyut Paudel ,&nbsp;Jostan Brown ,&nbsp;Priyanka Upadhyaya ,&nbsp;Atif Bilal Asad ,&nbsp;Safal Kshetri ,&nbsp;Joseph R. Davidson ,&nbsp;Cindy Grimm ,&nbsp;Ashley Thompson ,&nbsp;Bernardita Sallato ,&nbsp;Matthew D. Whiting ,&nbsp;Manoj Karkee","doi":"10.1016/j.compag.2025.110366","DOIUrl":"10.1016/j.compag.2025.110366","url":null,"abstract":"<div><div>Apple(<em>Malus domestica</em> Borkh.) trees are deciduous, shedding leaves each year. This process is preceded by a gradual change in leaf color from green to yellow as chlorophyll is degraded prior to abscission. The initiation and rate of this color change are affected by many factors including leaf nitrogen (N) concentration. We predict that leaf color during this transition may be indicative of the nitrogen status of apple trees. This study assesses a machine vision-based system for quantifying the change in leaf color and its correlation with leaf nitrogen content. An image dataset was collected in color and 3D over five weeks in the fall of 2021 and 2023 at a commercial orchard using a ground vehicle-based stereovision sensor. Trees in the foreground were segmented from the point cloud using color and depth thresholding methods. Then, to estimate the proportion of yellow leaves per canopy, the color information of the segmented canopy area was quantified using a custom-defined metric, “<em>yellowness index</em>” (a normalized ratio of yellow to green foliage in the tree) that varied from -1 to +1 (-1 being completely green and +1 being completely yellow). Both K-means-based methods and gradient boosting methods were used to estimate the <em>yellowness index</em>. The gradient boosting based method proposed in this study was better than the K-means-based method (both in terms of computational time and accuracy), achieving an <span><math><msup><mrow><mi>R</mi></mrow><mrow><mn>2</mn></mrow></msup></math></span> of 0.72 in estimating the <em>yellowness index</em>. The metric was able to capture the gradual color transition from green to yellow over the study duration. Trees with lower leaf nitrogen showed the color transition to yellow earlier than the trees with higher nitrogen. The onset of color transition during both years occurred during the 29th week post-full bloom (October 22 in 2021 and Nov 10 in 2023). This critical timing could be used for conducting nitrogen status analysis on apple trees using machine vision, enabling more precise and timely assessment of nutrient levels and facilitating targeted fertilization strategies in orchard management.</div></div>","PeriodicalId":50627,"journal":{"name":"Computers and Electronics in Agriculture","volume":"236 ","pages":"Article 110366"},"PeriodicalIF":7.7,"publicationDate":"2025-04-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143850528","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}