We've developed ILCD, a VQA model for crop diseases, integrating coattention, MUTAN, and BiBa. Our new CDwPK-VQA dataset includes multi-attribute info. ILCD achieves 86.06% accuracy on CDwPK-VQA, outperforming others.
Details: doi.org/10.34133/plantβ¦
We integrated hyperspectral & metabolomic data to identify salt-tolerant Medicago truncatula mutants. By combining data & using upper-level metabolomics, we achieved high efficiency & accuracy.
Details: doi.org/10.1016/j.plapβ¦
New study uses 3D RTM & LiDAR to analyze light in larch plantations. Crown volume key for APAR, while competition still impacts. Combining tech helps optimize forest structure & guide precise management. #forestry #ecology
Details: spj.science.org/doi/10.34133/pβ¦
New study uses YOLO v5, ResNet50, and DeepSORT models to analyze rice panicle traits via time - series images. Results show high accuracy in panicle counting and heading date estimation.
Details: spj.science.org/doi/10.34133/pβ¦
Boosting rice yield via enhanced photosynthesis! New study uses sun-induced chlorophyll fluorescence (SIF) to accurately estimate key traits like Vcmax and gs. π±π #agritech #photosynthesis #cropscience
Details: spj.science.org/doi/10.34133/pβ¦
New study proposes ALAEM algorithm to measure maize leaf orientation via RGB images. It shows leaf orientation varies by sowing density, row spacing, and genotype, impacting light interception.
Details: spj.science.org/doi/10.34133/pβ¦
We propose PLPNet for precise tomato leaf disease detection. It uses perceptual adaptive convolution, location reinforcement attention, and proximity feature aggregation to tackle challenges like soil interference. Details: spj.science.org/doi/10.34133/pβ¦
31.03.2025 14:48 β π 0 π 0 π¬ 0 π 0
New study: DC2Net detects Asian soybean rust early using hyperspectral imaging & deep learning. It combines deformable & dilated convolutions, achieving 96.73% accuracy. #CropDisease
Details: spj.science.org/doi/10.34133/pβ¦
New study uses AI & citizen science w/ smartphones to count coffee cherries on trees. Tested in Peru & Colombia, it shows promising results for scalable, low-cost crop monitoring in low-income regions. #AI #Coffee #CitizenScience
Details: spj.science.org/doi/10.34133/pβ¦
Hyperspectral imaging + machine learning = game-changing tool for analyzing pigments in Neopyropia! Fast, nondestructive, and accurate phenotyping of phycoerythrin, phycocyanin, allophycocyanin.
Details: spj.science.org/doi/10.34133/pβ¦
MG-YOLO: A novel detection algorithm for gray mold spores in precision ag. Combines Multi-head self-attention, BiFPN, and GhostCSP for high accuracy and speed. Achieves 0.983 accuracy in 0.009s/image, outperforming YOLOv5 by 6.8%.
Details: spj.science.org/doi/10.34133/pβ¦
PanicleNeRF uses smartphone videos to create 3D rice panicle models in fields. Combining SAM and YOLOv8, it achieves high segmentation accuracy and outperforms traditional methods. #Reconstruction
Details: spj.science.org/doi/10.34133/pβ¦
New VQA model for crop disease detection! ILCD uses co-attention, MUTAN, and bias balancing to identify disease stages. Achieves 86.06% accuracy on CDwPK-VQA dataset. Check it out: github.com/SdustZYP/ILCD-β¦
Details: spj.science.org/doi/10.34133/pοΏ½οΏ½οΏ½
New study improves LNC retrieval in Ginkgo trees using modified ratio indices & advanced BRF spectra methods. Results show enhanced accuracy & highlight potential for better nitrogen status assessment. #LeafNitrogen #RemoteSensing
Details: spj.science.org/doi/10.34133/pβ¦
New study develops LNA estimation model for wheat using UAV & hyperspectral data. Models consider vertical heterogeneity, improving accuracy. RF-LNASum model shows best results with 17.8% error.
Details: spj.science.org/doi/10.34133/pβ¦
New study uses field excavation & 3D digitization to analyze grapevine root systems, revealing genotype-specific water uptake. Excavation & in situ digitization are accurate, despite some fine root length underestimation. #Sustainableagriculture
Details: spj.science.org/doi/10.34133/pβ¦
Using drones & deep learning, we measured plant maturity, stand count, & height in fields. CNN-LSTM models improved maturity prediction. This tech offers accuracy & cost savings over traditional methods. #DeepLearning
Details: spj.science.org/doi/10.34133/pβ¦
We studied Phragmites australis & Typha orientalis to understand their canopy structure and solar radiation patterns. Key findings: layered solar radiation transmittance is more sensitive to canopy structure than pigments.
Details: spj.science.org/doi/10.34133/pβ¦
We propose using low-altitude aerial photography to create 3D point clouds and multispectral images of wheat plots. This helps extract dynamic digital phenotypes for genome-wide association studies. #Wheatplot #Image
Details: spj.science.org/doi/10.34133/pβ¦
Introducing CHCNet: A unified model for counting cereal crops like maize, rice, sorghum, and wheat using few-shot learning. It reduces labeling costs and enhances cross-crop generalization. Check it out at cerealcropnet.com
Details: spj.science.org/doi/10.34133/pβ¦
New AI model for plant disease diagnostics outperforms GPT-4! Uses 3 stages of image-text alignment to generate accurate phenotypic descriptions. Check it out: plantext.samlab.cn #ChartGPT
Details: spj.science.org/doi/10.34133/pβ¦
Our study explores nitrogen responsiveness in wheat using drone phenotyping & machine learning. We quantify traits, map genetic components, and classify varieties for optimized N use. #nitrogen
Details: spj.science.org/doi/10.34133/pβ¦
Exploring plant drought response? Chlorophyll fluorescence beats spectral reflectance in reliability. But using both methods together offers best insights. Study on tobacco & barley leaves shows the way! #ClimateChange #PlantScience
Details: spj.science.org/doi/10.34133/pβ¦
πΎ CSNet revolutionizes wheat breeding! πΎ Our count-supervised network uses quantity info to accurately count wheat ears without location data. Boosting yield & reducing costs! ππ± Learn more #AgTech #AI #FoodSecurity
Details: spj.science.org/doi/10.34133/pβ¦
Using AI and computer vision, we developed models to detect sorghum panicles and estimate grain numbers from smartphone images. Our Sorghum-Net model achieved a 17% error rate, paving the way for efficient crop yield estimation.
Details: spj.science.org/doi/10.34133/pβ¦
Studying root systems with germination papers offers high resolution but lacks natural conditions. Using FSPMs and RhizoVision, we analyzed Populus trichocarpa roots, finding thermal time correlation and varying 2D/3D accuracy.
Details: spj.science.org/doi/10.34133/pβ¦
We evaluated a novel 2-step workflow for automated root system architecture (RSA) reconstruction from MRI images. U-Net segmentation doubled reconstruction speed and increased root length in low CNR images.
Details: spj.science.org/doi/10.34133/pβ¦
Sugarcane, a major global food and bioenergy crop, is produced mostly in Brazil and India. Its breeding is slow and challenging due to its complex genome and long selection process.
Details: spj.science.org/doi/10.34133/pβ¦
Deep learning models for plant trait analysis need annotated datasets, which are labor-intensive to create. Our Helios 3D framework generates labeled synthetic plant images, easing data collection.
Details: spj.science.org/doi/10.34133/pβ¦
New study uses UAVs and deep learning to accurately detect maize tassels pre- and post-detasseling. Optimized models achieve 94.5% avg precision, with blocking strategies boosting accuracy to 98%. #agritech #maizehybridization
Details: spj.science.org/doi/10.34133/pβ¦
