Plant Phenomics

Cross-site TLS workflow estimates forest biomass without tree segmentation, using canopy structure metrics to improve carbon assessment across diverse East Asian ecosystems. #ForestBiomass #TLS #CarbonMonitoring Details:

HAE-Net introduces hierarchy-aware 3D panoptic segmentation for orchards, linking fruits to parent trees via multi-decoder architecture and hierarchical consistency loss. #OrchardRobotics #3DSegmentation #AgTech Details:

Hyperspectral imaging enables nondestructive early salt-tolerance screening during seed imbibition, identifying 500 genetic loci and key regulators of germination under stress. #SaltTolerance #Hyperspectral #PlantBreeding Details:

Few-shot segmentation with prototype-enhanced prompting achieves 99.57% accuracy for lettuce phenotyping in complex field environments, enabling high-throughput germplasm screening. #LettucePhenotyping #FewShotLearning #AgTech Details:

3D maize ear scanning pipeline using deep learning achieves R² up to 0.98 for kernel traits, enabling rapid, non-destructive phenotyping under varying phosphorus supply. #MaizePhenotyping #3DImaging #AgTech Details:

PAM fluorometry reveals maize cultivar differences in photochemical compensation points and photoprotective energy partitioning under varying nitrogen supply. #Maize #Photosynthesis #PlantPhysiology Details:

DINO-BollGX enables multi-temporal UAV monitoring of cotton boll-opening dynamics, optimizing harvest timing for mechanical picking through temporal stability indexing. #CottonPhenotyping #UAV #AgTech Details:

ZCAT enables zero-annotation cross-crop segmentation via human-in-the-loop pseudo-mask screening, cutting annotation costs and outperforming manual labels for wheat spikes. #PlantPhenomics #Segmentation #AgTech Details:

Unsupervised 3D multi-focus fusion framework reconstructs fully focused stomatal images, enabling accurate YOLO-based segmentation (mAP50 0.9937) and automated microscopic phenotyping. #Stomata #Microscopy #AgTech Details:

SAM-CLIP foundation models enable accurate thermal image segmentation for plant phenotyping with minimal supervision, achieving 97.54% mIoU on LadyBird dataset. #ThermalImaging #PlantPhenotyping #DeepLearning Details:

SAM-2 and GreenLab modeling track Arabidopsis leaf dynamics across environments, revealing genotype-specific growth strategies and novel QTLs for leaf emergence. #PlantPhenotyping #DeepLearning #QTLMapping Details:

GD-LNC integrates growth-stage embedding with dynamic feature fusion for UAV hyperspectral nitrogen monitoring in winter wheat, achieving R² of 0.88 and robust cross-stage adaptation. #PrecisionAg #Hyperspectral #WheatResearch Details:

ForestSeg3D improves individual tree segmentation from LiDAR point clouds using hierarchical semantic supervision and cross-task distillation, enhancing forest inventory accuracy. #ForestInventory #LiDAR #RemoteSensing Details:

Novel low-cost stereo depth camera method rapidly measures mean leaf inclination angle throughout crop canopies, enabling field-based evaluation of this key photosynthetic trait. #CropScience #Phenotyping #AgTech Details:

LUF-net integrates irradiance and exposure metadata via FiLM layers to correct UAV image colors, achieving <5.5% error across crops, altitudes, and lighting conditions. #UAVPhenotyping #ColorCorrection #AgTech Details:

Phenomics reveals developmental stage drives grassland responses to grazing, reshaping tiller traits and enabling precision management strategies. #GrasslandEcology #Phenomics #SustainableAg Details:

Ground-penetrating radar enables non-destructive sweet potato yield estimation by detecting underground storage roots, offering scalable field monitoring for root crops. #SweetPotato #GPR #AgTech Details:

High-throughput phenotyping and deep learning accelerate soybean breeding, enabling precise assessment of morphology, physiology, and agronomic traits at scale. #SoybeanBreeding #AI #PlantPhenotyping Details:

Deep-learning RGB phenotyping quantifies rice salt injury via shoot yellowing rate, enabling standardized genetic mapping and detection of novel salt-tolerance loci. #RiceScience #SaltTolerance #Phenotyping Details:

CitrusGS achieves high-fidelity 3D citrus reconstruction from just 15 X-ray projections using radiative Gaussian splatting, enabling automated phenotyping with R² > 0.944. #CitrusPhenotyping #3DReconstruction #AgTech Details: