Computed tomography enables non-destructive phenotyping of fruit internal structure but traditionally requires hundreds of projections, limiting throughput. Under sparse-view conditions, conventional and learning-based methods both suffer from streaking artifacts and regional distortions that degrade trait quantification. This study present CitrusGS, an integrated framework that achieves high-fidelity 3D reconstruction and precise morphological phenotyping of citrus fruit from only 15 projections using radiative 3D Gaussian splatting. Our method employs sparse-point initialization, optimized loss composite, and dual-stage pruning to suppress artifacts while preserving anatomically critical details with significantly higer convergence efficiency. In the citrus fruit datasets, CitrusGS achieves 29.78 dB PSNR and 0.870 SSIM, outperforming corresponding baseline method by 1.58 dB and 0.067 in SSIM, and enables automated extraction of ten external and internal phenotypic traits with R 2 larger than 0.944. Moreover, the framework shows initial zero-shot transferability across pathological citrus samples and additional horticultural specimens without retraining. By reconciling acquisition efficiency with anatomical fidelity using low-cost X-ray hardware, CitrusGS provides a promising framework for high-throughput, non-destructive phenotyping in breeding and grading applications.
Why it matches plant phenotyping methods柑橘果実の疎視野CT再構成法を開発・検証し、内部・外部形質を自動抽出するフェノタイピングが中心である。
abstractThis study present CitrusGS, an integrated framework that achieves high-fidelity 3D reconstruction and precise morphological phenotyping of citrus fruit from only 15 projections using radiative 3D Gaussian splatting.
Reproduction assets found保存済みの本文根拠を更新済みルールで再検証し、公開資産1件を確認しました。Code · publicOur codes are available at https://github.com/Petrichoror/CitrusGS .Open asset ↗Petrichoror/CitrusGSlines:325-387Code / dataset availability confirmedEurope PMC · Crossref · checked 5 Sept 2026
Plants are geometrically and topologically complex objects, and methods and devices that produce plant point clouds often miss parts due to self occlusions, making further analysis, such as phenotypic trait extraction or 3D reconstruction, difficult. We introduce A-Occ-Plant , a novel method for point cloud completion. The first novelty of our algorithm is converting point clouds into a set of images, which are then completed using 2D amodal segmentation. The images are then converted into a complete point cloud by using view-consistent Gaussian splats. The second novelty is the use of a coarse-to-fine hierarchical Transformer with cross-scale attention. The completed soft masks are fused into a continuous 3D density field using Gaussian splatting, removing the need for external pose estimation or fixed-size inputs. We introduce a synthetic dataset using a procedural model and a real-world plant reconstruction benchmark with artificially generated occlusions. We further benchmark A-Occ-Plant against representative 3D point-cloud completion methods, demonstrate that it recovers downstream phenotypic traits (leaf count, leaf angle, plant height), and show that it generalizes to another crops (soybean). A-Occ-Plant achieves a 264.8% improvement in LPIPS and an 8.3% gain in SSIM compared to the current state of the art, while using only 2.3% of the parameters and running 39.4× faster. We release our code at https://github.com/JaeLee18/PlantPhenomics_Occlusion.
Why it matches plant phenotyping methods植物の遮蔽点群を補完し、葉数・葉角度・草丈という表現型形質を復元する手法を開発しており、データセット作成とベンチマーク検証も中心的に行っている。
abstractWe introduce A-Occ-Plant , a novel method for point cloud completion.
Reproduction assets foundThe paper explicitly releases authors' code and sample data (inference code, sample data for reproducing results) via a Google Drive project download and a GitHub repository, both with explicit availability statements and public URLs.Code · publicThe full code and data at https://github.com/JaeLee18/PlantPhenomics_Occlusion .Open asset ↗JaeLee18/PlantPhenomics_Occlusionlines:386-410Code / dataset availability confirmedCrossref · Europe PMC · OpenAlex · checked 5 Sept 2026
Automated quantification of plant-level development from multi-plant greenhouse scenes requires separating individual plants from shared scene-level reconstructions and quantifying organ-level development, a challenge that single-plant acquisition workflows do not directly address. This study presents an end-to-end phenotyping pipeline built on 3D Gaussian Splatting (3DGS) and a post-reconstruction extraction framework, LCR-GS, designed to isolate plant instances from full greenhouse scenes without scene-specific model retraining. LCR-GS integrates zero-shot 2D cues with multi-view lifting, geometric clustering, and chromatic refinement to convert large scene-level reconstructions (~2M Gaussians) into compact per-plant subsets (~16K Gaussians). Experiments on greenhouse-grown muskmelon at the early vegetative stage demonstrate high plant-extraction precision (0.933) and strong organ-level instance segmentation (mean AP50 = 0.924). Plant height and leaf count are validated against manual measurements (height R² = 0.98, RMSE = 1.88 cm; leaf count R² = 0.86), whereas additional morphological traits, including leaf area, leaf area index, mean internode length, and stem node count, are reported as pipeline-derived descriptors for within-cohort comparison. By decoupling semantic inference from reconstruction, the pipeline reduces scene-scale data by over 99% and provides a practical route to derive compact per-plant 3D representations from multi-plant greenhouse imagery for downstream organ-level analysis.
Why it matches plant phenotyping methods3DGS画像から個体・器官を抽出し、植物形質を定量化するフェノタイピング手法の開発と検証が中心である。
abstractThis study presents an end-to-end phenotyping pipeline built on 3D Gaussian Splatting (3DGS) and a post-reconstruction extraction framework, LCR-GS, designed to isolate plant instances from full greenhouse scenes without scene-specific model retraining.
Reproduction assets foundThe paper's Data Availability Statement points to a public GitHub repository containing the muskmelon 3DGS phenotyping dataset (scenes, Gaussian-level plant/background annotations, and point-level organ labels) used in this study.Dataset · publicThe datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found below: https://github.com/bblabNTU/3dgs-muskmelon-phenotyping-dataset.Open asset ↗bblabNTU/3dgs-muskmelon-phenotyping-datasethtml-lines:485-547Code / dataset availability confirmedOpenAlex · Europe PMC · checked 5 Sept 2026
Efficient large-scale 3D reconstruction of orchard environments is essential for robotic inspection and precision agriculture, yet existing methods struggle with unstructured scenes, variable illumination, and computational bottlenecks. We propose InspectGaussian, a coarse-to-fine Gaussian reconstruction framework tailored for orchard inspection robots. The pipeline integrates an RGB-D-based data acquisition strategy using ORB-SLAM3, which is enhanced by a dense mapping module for robust large-scale pose estimation and point cloud generation. A divide-and-conquer strategy is then employed: individual plant views are extracted via a YOLO-World-based detection and 3D matching algorithm, followed by plant-specific reconstruction using an improved 3D Gaussian Splatting (3DGS) method incorporating depth regularization and region-aware refinement. Experimental results in citrus orchards demonstrate that InspectGaussian achieves 96% average precision and 93% recall in plant view extraction, while surpassing state-of-the-art methods in reconstruction fidelity (31.226 PSNR, 0.915 SSIM, 0.067 LPIPS) and point cloud accuracy (7 mm error). These results confirm its effectiveness in capturing fine structural and textural details while maintaining scalability and efficiency. This framework provides a practical solution for high-throughput, in-field plant phenotyping and lays the foundation for intelligent orchard monitoring and management.
Why it matches plant phenotyping methods植物個体の3D再構成とRGB-D・検出・Gaussian Splattingを統合した手法開発であり、植物の構造的形質取得を目的とするため、フェノタイピング手法が中心である。
abstractWe propose InspectGaussian, a coarse-to-fine Gaussian reconstruction framework tailored for orchard inspection robots.
Reproduction assets foundThe paper's authors explicitly state their analysis code is publicly available on GitHub. Phenotype datasets (RGB-D orchard image sequences, LiDAR point clouds, manual trait measurements) are only available upon request, so they do not qualify as public assets.Code · publicOur code are available at https://github.com/zlhzau/InspectGaussian.git .Open asset ↗zlhzau/InspectGaussianlines:489-515Code / dataset availability confirmedEurope PMC · OpenAlex · checked 15 Sept 2026
Fine-grained 3D phenotypic analysis of rice plays a vital role in rice breeding and yield estimation. However, a comprehensive rice data acquisition and segmentation pipeline is still lacking. While Neural Radiance Fields (NeRF) have shown impressive results in crop-level 3D reconstruction, their high sensitivity to data volume and camera viewpoints often leads to reconstruction failures for rice. In addition, the large-scale rice point clouds, coupled with heavy occlusion and visual similarity among grains, pose significant challenges for fine-grained trait extraction. To address the challenge of reconstructing rice point clouds under low-quality data conditions, we propose a novel method named Multi-Scale NeRF(MSNeRF). This method incorporates a structure-detail collaborative reconstruction mechanism and a dynamic initialization density scheduling strategy. Furthermore, we introduce a multimodal and multitask rice dataset (MMR) as a benchmark resource for future research. For rice point cloud segmentation, we develop Vision Rice Knowledge Graph Network(VRKGNet), which comprises an image segmentation module, a projection module, and a point cloud segmentation module enhanced with a Transformer to enlarge the receptive field. VRKGNet performs standalone point cloud segmentation and integrates image segmentation results from multiple viewpoints as prior knowledge to enhance semantic and instance-level segmentation. Extensive experiments demonstrate that MSNeRF achieves high-fidelity point cloud reconstruction with as few as 10 viewpoints. VRKGNet achieves superior rice plant segmentation with a semantic segmentation mIoU of 88.79% and an instance segmentation AP 25 of 84.55%, outperforming mainstream algorithms.
Why it matches plant phenotyping methods米の3D形質取得・再構成・分割を中核とする手法開発であり、データセット／ベンチマークも提供しているため、植物フェノタイピング手法文献に該当する。
abstractwe propose a novel method named Multi-Scale NeRF(MSNeRF)
Reproduction assets foundThe paper's authors explicitly state that the source code for MSNeRF and VRKGNet is publicly available on GitHub with testing scripts and test cases to reproduce the main results. The MMR dataset itself is only available upon request from the corresponding author, so it does not qualify as a public asset.Code · publicof Hefei Artificial Intelligence Breeding Accelerator Co. Ltd. ( NB2024005-02 ).
Data availability
The source code for the proposed methods, MSNeRF and VRKGNet, is publicly available on GitHub. The released repositories contain testing scripts and test cases used to reproduce the main results presented in this paper:
•
MSNeRF : https://github.com/qfwysw/MSNeRF.git
•
VRKGNet : https://github.com/qfwysw/VRKGNet.git
The datasets used in the experiments are available from the corresponding author upon reasonable request. For access or further inquiries, please contact the corresponding author.
Declaration of competing interest
The authors declare that they have no known competing financial iOpen asset ↗https://github.com/qfwysw/MSNeRF.gitlines:620-663Code · publicator Co. Ltd. ( NB2024005-02 ).
Data availability
The source code for the proposed methods, MSNeRF and VRKGNet, is publicly available on GitHub. The released repositories contain testing scripts and test cases used to reproduce the main results presented in this paper:
•
MSNeRF : https://github.com/qfwysw/MSNeRF.git
•
VRKGNet : https://github.com/qfwysw/VRKGNet.git
The datasets used in the experiments are available from the corresponding author upon reasonable request. For access or further inquiries, please contact the corresponding author.
Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could haveOpen asset ↗https://github.com/qfwysw/VRKGNet.gitlines:620-663Code / dataset availability confirmedOpenAlex · arXiv · checked 13 Sept 2026
Semantic reconstruction of agricultural scenes plays a vital role in tasks such as phenotyping and yield estimation. However, traditional approaches based on manual scanning or fixed camera setups remain a major bottleneck, while active-mapping methods based solely on occupancy grids are too coarse for accurate trait estimation. To address this gap, we propose an active 3D reconstruction framework for horticultural environments using a mobile manipulator. The system integrates OctoMap with 3D Gaussian Splatting to enable accurate and efficient target-aware mapping. A low-resolution OctoMap provides probabilistic occupancy information for informative viewpoint selection and collision-free planning, while 3D Gaussian Splatting leverages geometric, photometric, and semantic information to optimize 3D Gaussians for high-fidelity scene reconstruction. We further introduce a robust mapping strategy that mitigates semantic segmentation and depth noise, together with a background pruning method that reduces memory and computational cost. We validate our framework across simulated, laboratory, and real greenhouse scenes, showing consistent improvements across three state-of-the-art Gaussian Splatting backbones. In simulation, where ground-truth geometry is available, our approach outperforms occupancy-based mapping in both reconstruction accuracy and runtime efficiency: compared with a 0.01m-resolution OctoMap, it doubles the fruit-level F1 score under noisy conditions while achieving up to a threefold reduction in runtime. Beyond simulation, novel-view synthesis quality also improves consistently in laboratory and real greenhouse environments, with PSNR and mIoU improving by up to 1.5 dB and 18%, respectively. Finally, the reconstructed semantic maps enable fruit counting and volume estimation with accuracies approaching 80%.
Why it matches plant phenotyping methods園芸ロボット向けの3D再構成・能動マッピング手法を開発し、果実の計数・体積推定という植物形質の取得に適用・検証しているため、フェノタイピング手法が中心的です。
titleOctoSplat: Hybrid OctoMap-Gaussian Splatting for Active Semantic Mapping and Phenotyping with Horticultural Robots
Reproduction assets foundThe paper's supplementary material is hosted on the authors' public project page (jrcuaranv.github.io/octosplat), and the authors state that all code and data are publicly available. The SimSense repository is a third-party depth-sensor simulator tool, not a paper-specific asset.Code · publicAll code and data are publicly available to facilitate reproducibility.Open asset ↗lines:59-163Code / dataset availability confirmedEurope PMC · OpenAlex · checked 15 Sept 2026
Fruit growth has long been described using single- or double-sigmoid curves; however, these temporal models cannot fully capture the spatial heterogeneity that ultimately shapes a fruit. Here, we present a three-dimensional analysis pipeline that non-destructively tracks spatial fruit growth dynamics from field-collected imagery. Surface landmarks were drawn, and video recordings were taken throughout development for three pome fruits, apple ( Malus × domestica ), Japanese pear ( Pyrus pyrifolia ) and European pear ( Pyrus communis ), and two drupe fruits, peach ( Prunus persica ) and Japanese apricot ( Prunus mume ), to track their motion. Using 3D Gaussian splatting, we successfully reconstructed 3D models of the fruits, and the landmark displacement could be measured with high accuracy, with R 2 ≥ 0.98 when compared to manual recordings. We found a common spatial growth gradient in the longitudinal growth shared in the pomes and drupes of the Rosaceae; proximal (stem-end) regions exhibited more pronounced growth than the distal (stylar) end. An exception was found in European pear 'Bartlett,' which showed relatively vigorous growth in the distal region, explaining its distinct shape with expanded distal end. Transverse expansion varied far less than longitudinal expansion, with a possible association with initial fruit morphology. Inter-fruit growth variability peaked in the fastest-growing regions, particularly in the distal area of the European pear, highlighting the link between growth vigor and phenotypic variance. These results provide foundational insights into the developmental dynamics of both pome and drupe fruits of the Rosaceae family, contributing to the optimization of fruit size, shape, and uniformity.
Why it matches plant phenotyping methods3D画像解析パイプラインと3D Gaussian Splattingを用いて果実の空間的成長を非破壊計測し、手動記録との精度比較で検証しているため、植物表現型取得法が中心である。
abstractHere, we present a three-dimensional analysis pipeline that non-destructively tracks spatial fruit growth dynamics from field-collected imagery.
Reproduction assets foundThe authors deposited a subset of the 3DGS-reconstructed fruit models (the paper's phenotyping outputs) on Figshare with a public DOI; additional data only on request. No author analysis code or raw imagery deposit is stated.Dataset · publicFootnotes
Appendix A
Supplementary data to this article can be found online at https://doi.org/10.1016/j.plaphe.2026.100166 .
Appendix A.
Supplementary data
The following is the Supplementary data to this article:
Multimedia component 1
Multimedia component 1
Data availability
A subset of the generated 3D models is available at https://doi.org/10.6084/m9.figshare.30854579 , where the quality of the 3DGS reconstructions and the marking/measurement procedure can be examined. Additional data may be provided upon reasonable request to the corresponding author.
ReferencesOpen asset ↗figshare · 10.6084/m9.figshare.30854579lines:151-171Code / dataset availability confirmedarXiv · OpenAlex · checked 15 Sept 2026
AppleCottonPearField / plotNeRF / 3D Gaussian SplattingFruitCounting2D/3D reconstructionSegmentation
Rigorous crop counting is crucial for effective agricultural management and informed intervention strategies. However, in outdoor field environments, partial occlusions combined with inherent ambiguity in distinguishing clustered crops from individual viewpoints poses an immense challenge for image-based segmentation methods. To address these problems, we introduce a novel crop counting framework designed for exact enumeration via 3D instance segmentation. Our approach utilizes 2D images captured from multiple viewpoints and associates independent instance masks for neural radiance field (NeRF) view synthesis. We introduce crop visibility and mask consistency scores, which are incorporated alongside 3D information from a NeRF model. This results in an effective segmentation of crop instances in 3D and highly-accurate crop counts. Furthermore, our method eliminates the dependence on crop-specific parameter tuning. We validate our framework on three agricultural datasets consisting of cotton bolls, apples, and pears, and demonstrate consistent counting performance despite major variations in crop color, shape, and size. A comparative analysis against the state of the art highlights superior performance on crop counting tasks. Lastly, we contribute a cotton plant dataset to advance further research on this topic.
Why it matches plant phenotyping methodsNeRFと3Dインスタンスセグメンテーションを用いて作物個体・器官数を推定する画像ベース表現型計測手法を開発・検証しており、方法が研究の中心である。
abstractwe introduce a novel crop counting framework designed for exact enumeration via 3D instance segmentation.
Reproduction assets foundThe paper contributes a public infield cotton plant dataset (8 plants, ~150 iPhone images each, ground-truth boll counts, SAM instance masks) and states that source code, dataset, and multimedia are available at the authors' public project page, which is an allowed URL. The spectacularai GitHub URL is a generic third-pDataset · publicthat incorporates crop visibility
and mask consistency, enabling robustness against occlusions and annotation
discrepancies.
•
We release a public infield cotton plant dataset designed for 3D
rendering and cotton boll counting tasks.
The source code, dataset, and multimedia material associated with this project
can be found at
https://robotic-vision-lab.github.io/cropnerf .
II Related Work
II-A Image-Based Techniques
Image-based methods typically employ object detection to identify crops within
images. For example, Chen et al. [ 4 ] utilized multiple
convolutional neural networks (CNNs) to map input images to total fruit counts.
Similarly, Häni et al. [ 5 ] formulated crop counting as a
multOpen asset ↗lines:108-187Code / dataset availability confirmedarXiv · OpenAlex · checked 13 Sept 2026
Automating tasks in orchards is challenging because of the large amount of variation in the environment and occlusions. One of the challenges is apple pose estimation, where key points, such as the calyx, are often occluded. Recently developed pose estimation methods no longer rely on these key points, but still require them for annotations, making annotating challenging and time-consuming. Due to the abovementioned occlusions, there can be conflicting and missing annotations of the same fruit between different images. Novel 3D reconstruction methods can be used to simplify annotating and enlarge datasets. We propose a novel pipeline consisting of 3D Gaussian Splatting to reconstruct an orchard scene, simplified annotations, automated projection of the annotations to images, and the training and evaluation of a pose estimation method. Using our pipeline, 105 manual annotations were required to obtain 28,191 training labels, a reduction of 99.6%. Experimental results indicated that training with labels of fruits that are $\leq95\%$ occluded resulted in the best performance, with a neutral F1 score of 0.927 on the original images and 0.970 on the rendered images. Adjusting the size of the training dataset had small effects on the model performance in terms of F1 score and pose estimation accuracy. It was found that the least occluded fruits had the best position estimation, which worsened as the fruits became more occluded. It was also found that the tested pose estimation method was unable to correctly learn the orientation estimation of apples.
Why it matches plant phenotyping methodsリンゴの姿勢推定アノテーションを大幅に効率化する3D再構成・自動ラベル投影パイプラインを開発し、姿勢推定性能も評価しているため、植物フェノタイピング手法が中心である。
abstractWe propose a novel pipeline consisting of 3D Gaussian Splatting to reconstruct an orchard scene, simplified annotations, automated projection of the annotations to images, and the training and evaluation of a pose estimation method.
Reproduction assets foundThe paper explicitly provides two paper-specific public assets: the authors' phenotyping/pose-estimation pipeline code on GitHub and the collected apple orchard image dataset on a 4TU DOI. Both are directly used for the paper's measurements and analysis.Dataset · publicIn total, 367 images were collected. The dataset is available at https://doi.org/10.4121/976c94f2-028f-4291-adfd-20eb82b0f647Open asset ↗10.4121/976c94f2-028f-4291-adfd-20eb82b0f647lines:92-108Code / dataset availability confirmedCrossref · OpenAlex · Europe PMC · checked 6 Sept 2026
RiceWheatField / plotMesh / voxelNeRF / 3D Gaussian SplattingLiDAR / point cloudPanicle / ear / spikeLeafRootSeed / grain
Advanced plant phenotyping technologies are vital for trait improvement and accelerating intelligent breeding. Due to the species diversity of plants, existing methods heavily rely on large-scale high-precision manually annotated data. For self-occluded objects at the grain level, unsupervised methods often prove ineffective. This study proposes IPENS, an interactive unsupervised multi-target point cloud extraction method. It utilizes radiance field information to lift 2D masks, segmented by SAM2 (Segment Anything Model 2), into 3D space for target point cloud extraction. A multi-target collaborative optimization strategy addresses the challenge of segmenting multiple targets from a single interaction. On a rice dataset, IPENS achieves a grain-level segmentation mean Intersection over Union (mIoU) of 63.72%. For phenotypic trait estimation, it achieves a grain voxel volume coefficient of determination R 2 = 0.7697 (Root Mean Square Error, RMSE = 0.0025), leaf surface area R 2 = 0.84 (RMSE = 18.93), and leaf length and width prediction accuracies of R 2 = 0.97 and R 2 = 0.87 (RMSE = 1.49 and 0.21). On a wheat dataset, IPENS further improves segmentation performance to a mIoU of 89.68%, with exceptional phenotypic estimation results: panicle voxel volume R 2 = 0.9956 (RMSE = 0.0055), leaf surface area R 2 = 1.00 (RMSE = 0.67), and leaf length and width predictions reaching R 2 = 0.99 and R 2 = 0.92 (RMSE = 0.23 and 0.15). Without requiring annotated data, IPENS rapidly extracts grain-level point clouds for multiple targets within three minutes using single-round image interactions. These features make IPENS a high-quality, non-invasive phenotypic extraction solution for rice and wheat, offering significant potential to enhance intelligent breeding.
Why it matches plant phenotyping methods植物形質抽出のためのNeRF-SAM2融合手法を開発し、作物データセットで分割性能と形質推定精度を検証しているため、方法開発・検証が中心である。
abstractThis study proposes IPENS, an interactive unsupervised multi-target point cloud extraction method.
Reproduction assets foundThe paper's analysis code is publicly available on GitHub. The rice/wheat MMR/MMW phenotype datasets (multi-view images, point clouds, annotations) are only available upon reasonable request, so they are not public.Code · publicCode is available at https://github.com/Vincent-Songwentao/IPENS-Code.git .Open asset ↗https://github.com/Vincent-Songwentao/IPENS-Code.gitlines:472-496Code / dataset availability confirmedOpenAlex · Europe PMC · checked 15 Sept 2026
Rapeseed / canolaNeRF / 3D Gaussian SplattingLiDAR / point cloudFruitCountingSegmentationFruit / seed / panicle traits
Accurate analysis of plant phenotypic traits is crucial for crop breeding and precision agriculture. This study proposes a lightweight semantic segmentation model named KAN-GLNet (Kolmogorov-Arnold Network with Global-Local Feature Modulation), based on an enhanced PointNet++ architecture and integrated with an optimized Density-Based Spatial Clustering of Applications with Noise (DBSCAN) algorithm, to achieve high-precision segmentation and automatic counting of canola siliques. A multi-view point cloud acquisition platform was built, and high-fidelity canola point clouds were reconstructed using Neural Radiance Fields (NeRF) technology. The proposed model includes three key modules: Reverse Bottleneck Kolmogorov-Arnold Network Convolution, a Global-Local Feature Modulation (GLFN) block, and a contrastive learning-based normalization module called ContraNorm. KAN-GLNet contains only 5.72M parameters and achieves 94.50% mIoU, 96.72% mAcc, and 97.77% OAcc in semantic segmentation tasks, outperforming all baseline models. In addition, the DBSCAN workflow was optimized, achieving a counting accuracy of 97.45% in the instance segmentation task. This method achieves an excellent balance between segmentation accuracy and model complexity, providing an efficient solution for high-throughput plant phenotyping. The code and dataset have been made publicly available at: https://anonymous.4open.science/r/KAN-GLNet-6432/.
Why it matches plant phenotyping methodsカノーラ莢のセグメンテーションと自動計数という植物形質抽出手法を、3D点群取得基盤・NeRF再構成・新規モデル・DBSCANワークフローとして開発・評価しており、植物フェノタイピング手法が中心である。
abstractThis study proposes a lightweight semantic segmentation model named KAN-GLNet (Kolmogorov-Arnold Network with Global-Local Feature Modulation), based on an enhanced PointNet++ architecture and integrated with an optimized Density-Based Spatial Clustering of Applications with Noise (DBSCAN) algorithm, to achieve high-precision segmentation and automatic counting of canola siliques.
Reproduction assets foundThe authors explicitly state that their curated code and dataset (canola silique point cloud phenotyping data and KAN-GLNet analysis code) are publicly available at an anonymous.4open.science repository, which is an allowed URL.Code · publicDBSCAN workflow was optimized, achieving a counting accuracy of 97.45% in the instance segmentation task. This method achieves an excellent balance between segmentation accuracy and model complexity, providing an efficient solution for high-throughput plant phenotyping. The code and dataset have been made publicly available at: https://anonymous.4open.science/r/KAN-GLNet-6432/ .
http://dx.doi.org/10.13039/501100001809 National Natural Science Foundation of China
32301762
Liu Jie
This project is supported by National Natural Science Foundation of China, grant number 32301762.
pmc-status-qastatus 0 pmc-status-live yes pmc-status-embargo no pmc-status-released yes pmc-prop-open-access yes pmc-pOpen asset ↗anonymous.4open.science/r/KAN-GLNet-6432lines:1-65Code / dataset availability confirmedEurope PMC · bioRxiv · OpenAlex · checked 15 Sept 2026
Fruit growth has long been described using single- or double-sigmoid curves; however, these temporal models cannot fully capture the spatial heterogeneity that ultimately shapes a fruit. Here, we present a three-dimensional analysis pipeline that non-destructively tracks spatial fruit growth dynamics from field-collected imaginary. Surface landmarks were drawn, and video recordings were taken throughout development for three pome fruits, apple ( Malus × domestica ), Japanese pear ( Pyrus pyrifolia ) and European pear ( Pyrus communis ), and two drupe fruits, peach ( Prunus persica ) and Japanese apricot ( Prunus mume ), to track their motion. Using 3D Gaussian Splatting, we successfully reconstructed 3D models of the fruits, and the landmark displacement could be measured with high accuracy, with R 2 ≥ 0.98 when compared to manual recordings. We found a common spatial growth gradient in the longitudinal growth shared in the pomes and drupes of the Rosaceae; proximal (stem-end) regions exhibited more pronounced growth than the distal (stylar) end. An exception was found in European pear ‘Bartlett,’ which showed relatively vigorous growth in the distal region, explaining its distinct shape with expanded distal end. Transverse expansion varied far less than longitudinal expansion, with a possible association with initial fruit morphology. Inter-fruit growth variability peaked in the fastest-growing regions, particularly in the distal area of the European pear, highlighting the link between growth vigor and phenotypic variance. These results provide foundational insights into the developmental dynamics of both pome and drupe fruits of the Rosaceae family, contributing to the optimization of fruit size, shape, and uniformity.
Why it matches plant phenotyping methods3D Gaussian Splattingを用いて果実の3D再構成と空間的成長形質の非破壊計測パイプラインを開発し、手動記録との精度比較で検証しているため、フェノタイピング手法が中心である。
abstractHere, we present a three-dimensional analysis pipeline that non-destructively tracks spatial fruit growth dynamics
Reproduction assets foundThe paper's data availability statement deposits a subset of the generated 3DGS fruit reconstruction models (the paper's phenotyping outputs) on Figshare with a public DOI; additional data is request-only. No author analysis code is explicitly deposited.Dataset · publicFootnotes
Appendix A
Supplementary data to this article can be found online at https://doi.org/10.1016/j.plaphe.2026.100166 .
Appendix A.
Supplementary data
The following is the Supplementary data to this article:
Multimedia component 1
Multimedia component 1
Data availability
A subset of the generated 3D models is available at https://doi.org/10.6084/m9.figshare.30854579 , where the quality of the 3DGS reconstructions and the marking/measurement procedure can be examined. Additional data may be provided upon reasonable request to the corresponding author.
ReferencesOpen asset ↗figshare · 10.6084/m9.figshare.30854579lines:151-171Code / dataset availability confirmedOpenAlex · Europe PMC · checked 13 Sept 2026
Accurately reconstructing tree skeletons from multi-view images is challenging. While most existing works use skeletonization from 3D point clouds, thin branches with low-texture contrast often involve multi-view stereo (MVS) to produce noisy and fragmented point clouds, which break branch connectivity. Leveraging the recent development in accurate mask extraction from images, we introduce a mask-guided graph optimization framework that estimates a 3D skeleton directly from multi-view segmentation masks, bypassing the reliance on point cloud quality. In our method, a skeleton is modeled as a graph whose nodes store positions and radii while its adjacency matrix encodes branch connectivity. We use 3D Gaussian splatting (3DGS) to render silhouettes of the graph and directly optimize the nodes and the adjacency matrix to fit given multi-view silhouettes in a differentiable manner. Furthermore, we use a minimum spanning tree (MST) algorithm during the optimization loop to regularize the graph to a tree structure. Experiments on synthetic and real-world plants show consistent improvements in completeness and structural accuracy over existing point-cloud-based and heuristic baseline methods.
Why it matches plant phenotyping methods植物のマルチビュー画像から樹木の3D骨格・枝構造を推定する計算手法を開発し、実植物で既存法と比較検証しているため、植物形態フェノタイピング手法が中心である。
abstractwe introduce a mask-guided graph optimization framework that estimates a 3D skeleton directly from multi-view segmentation masks
Reproduction assets foundThe paper's authors explicitly state their implementation is publicly available on GitHub, which is the paper-specific computational analysis code for the mask-guided tree skeleton extraction method.Code · publicOur implementation is available in the public GitHub repository ( https://github.com/huntorochi/Masks-to-Skeleton , accessed on 18 May 2025).Open asset ↗huntorochi/Masks-to-Skeletonlines:27-41Code / dataset availability confirmedOpenAlex · arXiv · checked 14 Sept 2026
This paper presents a NeRF-based framework for point cloud (PCD) reconstruction, specifically designed for indoor high-throughput plant phenotyping facilities. Traditional NeRF-based reconstruction methods require cameras to move around stationary objects, but this approach is impractical for high-throughput environments where objects are rapidly imaged while moving on conveyors or rotating pedestals. To address this limitation, we develop a variant of NeRF-based PCD reconstruction that uses a single stationary camera to capture images as the object rotates on a pedestal. Our workflow comprises COLMAP-based pose estimation, a straightforward pose transformation to simulate camera movement, and subsequent standard NeRF training. A defined Region of Interest (ROI) excludes irrelevant scene data, enabling the generation of high-resolution point clouds (10M points). Experimental results demonstrate excellent reconstruction fidelity, with precision-recall analyses yielding an F-score close to 100.00 across all evaluated plant objects. Although pose estimation remains computationally intensive with a stationary camera setup, overall training and reconstruction times are competitive, validating the method's feasibility for practical high-throughput indoor phenotyping applications. Our findings indicate that high-quality NeRF-based 3D reconstructions are achievable using a stationary camera, eliminating the need for complex camera motion or costly imaging equipment. This approach is especially beneficial when employing expensive and delicate instruments, such as hyperspectral cameras, for 3D plant phenotyping. Future work will focus on optimizing pose estimation techniques and further streamlining the methodology to facilitate seamless integration into automated, high-throughput 3D phenotyping pipelines.
Why it matches plant phenotyping methods植物フェノタイピング施設向けに、固定カメラ画像からNeRFで植物の3D点群を再構成する手法を開発・評価しており、表現型取得が研究の中心である。
abstractThis paper presents a NeRF-based framework for point cloud (PCD) reconstruction, specifically designed for indoor high-throughput plant phenotyping facilities.
Reproduction assets foundThe paper explicitly releases its full SC-NeRF dataset (raw 4K videos, frames, COLMAP poses, NeRF checkpoints, and final 10M-point clouds for six plant/produce objects) on Hugging Face, and states that all datasets and the authors' code are available at the project page. Both are paper-specific, public, and actionable.Code · publicd delicate instruments, such as hyperspectral cameras, for 3D plant phenotyping. Future work will focus on optimizing pose estimation techniques and further streamlining the methodology to facilitate seamless integration into automated, high-throughput 3D phenotyping pipelines. We provide all datasets and our code, available at https://baskargroup.github.io/SC-NeRF/
Figure 1 : Schematic of the stationary camera imaging system for NeRF-based point cloud reconstruction in high-throughput plant phenotyping. In this setup, each plant is conveyed to a rotating turntable marked against a matte black background. Over a full 30-second rotation, a tripod-mounted stationary camera captures high-resoOpen asset ↗lines:1-53Code / dataset availability confirmedOpenAlex · checked 15 Sept 2026
Field / plotNeRF / 3D Gaussian SplattingLiDAR / point cloudStem / branchMorphology / geometry measurement2D/3D reconstructionArchitecture / morphology / geometry
Forest mapping provides critical observational data needed to understand the dynamics of forest environments. Notably, tree diameter at breast height (DBH) is a metric used to estimate forest biomass and carbon dioxide (CO2) sequestration. Manual methods of forest mapping are labor intensive and time consuming, a bottleneck for large-scale mapping efforts. Automated mapping relies on acquiring dense forest reconstructions, typically in the form of point clouds. Terrestrial laser scanning (TLS) and mobile laser scanning (MLS) generate point clouds using expensive LiDAR sensing and have been used successfully to estimate tree diameter. Neural radiance fields (NeRFs) are an emergent technology enabling photorealistic, vision-based reconstruction by training a neural network on a sparse set of input views. In this paper, we present a comparison of MLS and NeRF forest reconstructions for the purpose of trunk diameter estimation in a mixed-evergreen Redwood forest. In addition, we propose an improved DBH-estimation method using convex-hull modeling. Using this approach, we achieved 1.68 cm RMSE (2.81%), which consistently outperformed standard cylinder modeling approaches.
Why it matches plant phenotyping methods森林内の樹木DBHという個体形態形質を、NeRF・MLS再構成と凸包モデルで推定し、手法比較と精度評価を行っているため、植物形質取得法が中心です。
abstractIn this paper, we present a comparison of MLS and NeRF forest reconstructions for the purpose of trunk diameter estimation in a mixed-evergreen Redwood forest.
Reproduction assets foundThe authors explicitly state that their forest datasets (SLAM and NeRF reconstructions, imagery) and TreeTool modeling code contributions are freely available on their public GitHub repository, which is listed in the allowed URLs.Code · publicAR-inertial SLAM with regards to DBH estimation accuracy.
• Improved DBH estimation accuracy via a trunk modeling approach using convex-hull
and density-based filtering methods.
• Open-source modeling code and forest datasets, including SLAM and NeRF recon-
structions of a mixed-evergreen Redwood forest, are freely available at
https://github.com/harelab-ucsc/RedwoodNeRF (accessed on 7 January 2025).
2. Theoretical Background
2.1. The SLAM Approach
The SLAM problem can be broken into two tasks: building a map of the environment
and simultaneously estimating the robot’s trajectory within that map. More specifically,Open asset ↗harelab-ucsc/RedwoodNeRFpdf-raw-page:2 lines:1-50Code / dataset availability confirmedOpenAlex · Europe PMC · checked 15 Sept 2026
BACKGROUND: The reconstruction of 3-dimensional (3D) plant models can offer advantages over traditional 2-dimensional approaches by more accurately capturing the complex structure and characteristics of different crops. Conventional 3D reconstruction techniques often produce sparse or noisy representations of plants using software or are expensive to capture in hardware. Recently, view synthesis models have been developed that can generate detailed 3D scenes, and even 3D models, from only RGB images and camera poses. These models offer unparalleled accuracy but are currently data hungry, requiring large numbers of views with very accurate camera calibration. RESULTS: In this study, we present a view synthesis dataset comprising 20 individual wheat plants captured across 6 different time frames over a 15-week growth period. We develop a camera capture system using 2 robotic arms combined with a turntable, controlled by a re-deployable and flexible image capture framework. We trained each plant instance using two recent view synthesis models: 3D Gaussian splatting (3DGS) and neural radiance fields (NeRF). Our results show that both 3DGS and NeRF produce high-fidelity reconstructed images of a plant subject from views not captured in the initial training sets. We also show that these approaches can be used to generate accurate 3D representations of these plants as point clouds, with 0.74-mm and 1.43-mm average accuracy compared with a handheld scanner for 3DGS and NeRF, respectively. CONCLUSION: We believe that these new methods will be transformative in the field of 3D plant phenotyping, plant reconstruction, and active vision. To further this cause, we release all robot configuration and control software, alongside our extensive multiview dataset. We also release all scripts necessary to train both 3DGS and NeRF, all trained models data, and final 3D point cloud representations. Our dataset can be accessed via https://plantimages.nottingham.ac.uk/ or https://https://doi.org/10.5524/102661. Our software can be accessed via https://github.com/Lewis-Stuart-11/3D-Plant-View-Synthesis.
Why it matches plant phenotyping methods3D植物表現型取得のための撮影システム、再構成手法、データセットを開発し、スキャナとの精度比較で検証しているため、方法が中心的である。
abstractWe develop a camera capture system using 2 robotic arms combined with a turntable, controlled by a re-deployable and flexible image capture framework.
Reproduction assets foundThe paper releases its wheat plant multiview image dataset (via plantimages.nottingham.ac.uk and GigaDB DOI 10.5524/102661), its authors' analysis/capture codebase on GitHub (3D-Plant-View-Synthesis), a Software Heritage archive of that code, and a DOME-ML registry annotation. All are paper-specific, public, and have作者Code · publicruction output across all plants. We hope that our study will provide opportunities for researchers exploring new and improved 3D phenotyping algorithms, 3D reconstruction and view synthesis research, and active vision systems.
Availability of Source Code and Requirements
Project name: 3D Plant View Synthesis:
Project homepage: https://github.com/Lewis-Stuart-11/3D-Plant-View-Synthesis [ 13 ]
Operating system(s): Windows, Ubuntu
Programming language: Python (>=3.8)
License: Apache 2.0
Any restrictions to use by nonacademics: None
Our code has also been archived in Software Heritage [ 66 ]. Functionality, such as Robotic View Capturing, 3DGS to Point Cloud, and our UR5 Configs files, are storOpen asset ↗GitHub · Lewis-Stuart-11/3D-Plant-View-Synthesislines:663-695Code / dataset availability confirmedarXiv · OpenAlex · checked 13 Sept 2026
Forest mapping provides critical observational data needed to understand the dynamics of forest environments. Notably, tree diameter at breast height (DBH) is a metric used to estimate forest biomass and carbon dioxide sequestration. Manual methods of forest mapping are labor intensive and time consuming, a bottleneck for large-scale mapping efforts. Automated mapping relies on acquiring dense forest reconstructions, typically in the form of point clouds. Terrestrial laser scanning (TLS) and mobile laser scanning (MLS) generate point clouds using expensive LiDAR sensing, and have been used successfully to estimate tree diameter. Neural radiance fields (NeRFs) are an emergent technology enabling photorealistic, vision-based reconstruction by training a neural network on a sparse set of input views. In this paper, we present a comparison of MLS and NeRF forest reconstructions for the purpose of trunk diameter estimation in a mixed-evergreen Redwood forest. In addition, we propose an improved DBH-estimation method using convex-hull modeling. Using this approach, we achieved 1.68 cm RMSE, which consistently outperformed standard cylinder modeling approaches. Our code contributions and forest datasets are freely available at https://github.com/harelab-ucsc/RedwoodNeRF.
Why it matches plant phenotyping methodsNeRFおよびMLSによる森林再構成から樹木DBHを推定し、凸包モデルによる推定法を提案・比較検証しているため、植物形質取得手法が中心です。
abstractIn this paper, we present a comparison of MLS and NeRF forest reconstructions for the purpose of trunk diameter estimation in a mixed-evergreen Redwood forest.
Reproduction assets foundThe authors explicitly state their code contributions and forest datasets (SLAM and NeRF reconstructions used for DBH estimation) are freely available in a public GitHub repository. The other URLs are a cited third-party tool (NeRFCapture) and a background reference (USDA aerial survey), neither of which is a paper-ownDataset · publicOur code contributions and forest datasets are freely available at https://github.com/harelab-ucsc/RedwoodNeRF .Open asset ↗harelab-ucsc/RedwoodNeRFlines:1-52