Smart Agriculture and AI

Vision Normalizing Flows for the probability-informed detection of banana diseases from in-field images

The paper develops a vision normalizing flow model to detect banana diseases from in-field images with high accuracy, using pre-trained embeddings and probabilistic density estimation.

A Novel Phenotyping Approach for Reconciling Precision and Variance in Disease Severity Estimates from High-resolution Imaging

The paper develops an imaging approach to improve disease severity estimation in wheat by combining high-resolution imaging and focus bracketing, reducing measurement uncertainty and enabling efficient sampling.

Petal to the metal: The slow road to automating large-scale phenology labeling for herbarium specimens

The paper presents a machine learning pipeline for automating flower detection on herbarium specimens, resulting in 11.1 million labeled records but highlighting the need for full metadata digitization.

Segment Any Plant (SAP): Foundation-Model Segmentation for Plant Time-Series Phenotyping

SAP provides a transferable, training-free framework for automated segmentation of plant time-series imagery across species and conditions.

Text guidance is powerful but prompt-sensitive for weakly-supervised leaf symptom segmentation

The paper evaluates text-guided segmentation using SAM3 for weakly supervised leaf symptom segmentation, finding text guidance matches or outperforms conventional methods but is prompt-sensitive, with hybrid approaches improving robustness.

AI-Powered Acoustic Surveillance for Early Detection of Calf Respiratory Disease

This paper develops an AI-powered acoustic monitoring system to detect coughs in calves, achieving 92% accuracy and enabling early detection of respiratory disease 1-2 days before clinical signs.

AGIcam: An open-source IoT-based camera system for automated in-field phenotyping and yield prediction

AGIcam provides an IoT-based camera system for continuous in-field plant phenotyping and yield prediction using time-series imagery and machine learning models.

An Expert-Informed Synthetic Animal Data Generator: A Physiology-Consistent Generative Framework for High-Fidelity Animal Digital Twins

This paper introduces EICD, a framework that generates physiology-consistent synthetic animal data by embedding mechanistic biological principles, achieving high fidelity and reducing physiological hallucinations.

Image-based leaf SPAD value and chlorophyll measurement using a mobile phone: enabling accessible and sustainable crop management

This paper develops a mobile phone-based app to estimate leaf SPAD and chlorophyll values in rice with accuracy comparable to lab methods and commercial tools.

Phenotypic scoring of Canola Blackleg severity using machine learning image analysis

This paper introduces a deep learning model to score Canola Blackleg severity from images, showing it is more consistent than expert raters and maintains similar heritability for breeding purposes.

PlantCV v4: Image analysis software for high-throughput plant phenotyping

PlantCV v4 improves usability and functionality for plant phenotyping by adding new image types and tutorials for non-coding researchers.

Samplify: A versatile tool for image-based segmentation and annotation of seed abortion phenotypes

Samplify automates seed segmentation and classification in Arabidopsis, improving efficiency and reducing user bias in seed abortion phenotyping.

Towards an AI-based knowledge assistant for goat farmers based on Retrieval-Augmented Generation

This paper develops an AI-based knowledge assistant for goat farmers using Retrieval-Augmented Generation to improve LLM performance on structured knowledge tasks.

Strategic Segmentation: A Nash Equilibrium based Approach for Weed Segmentation in Agricultural Fields

This paper develops a deep learning method using game theory to improve weed segmentation in agricultural fields by finding a Nash equilibrium to reduce false positives.

Automated detection and quantification of two-spotted spider mite life stages using computer vision for high-throughput in vitro assays

This paper develops a computer vision pipeline for automated detection of two-spotted spider mite life stages, achieving high accuracy in fecundity estimation but lower accuracy in mortality assessment.

Deep Learning for Sorghum Yield Forecasting using Uncrewed Aerial Systems and Lab-Derived Imagery

This paper uses deep learning models to detect sorghum panicles from aerial and lab images, improving yield prediction through machine learning regression.

Leveraging High-dimensional Seed Colorspace in Phenomic Selection Models for Herbaceous Perennial Crops

The paper uses seed color and morphology data from image scans to predict germination traits in perennial crops, showing improved prediction accuracy when combining HSV color features with morphology.

LeafyResNet: Fusarium Wilt Detection in Lettuce Using UAV RGB Imaging and Advanced Deep Learning Model

This paper develops LeafyResNet, a deep learning model for detecting Fusarium wilt in lettuce using UAV RGB imaging, achieving high accuracy and precision.

pyRootHair: Machine Learning Accelerated Software for High-Throughput Phenotyping of Plant Root Hair Traits

pyRootHair automates root hair trait quantification from plant root images, revealing significant variation across wheat cultivars and supporting high-throughput analysis.

Impact of cross-validation designs on cattle behavior prediction using machine learning and deep learning models with tri-axial accelerometer data

Leave-cow-out validation lowered cattle behavior classification accuracy to 0.72 to 0.82, versus 0.9 to 0.96 under random hold-out splits that can place one animal's records in both sets.

Integrating Load-Cell Lysimetry and Machine Learning for Prediction of Daily Plant Transpiration

This study uses machine learning models trained on gravimetric and environmental data to predict daily plant transpiration with high accuracy.

Deep learning from videography as a tool for measuring E. coli infection in poultry

This study uses deep learning-based tracking to detect behavioral changes in infected poultry, showing reduced activity in E. coli-infected chickens compared to controls.

Multi-Model Machine Learning for Automated Identification of Rice Diseases Using Leaf Image Data

This paper develops a hybrid deep-machine learning system for automated rice disease detection using leaf images, achieving high accuracy and efficiency.

Synergistic 3D, multispectral, and thermal image analysis via supervised machine learning for improved detection of root rot symptoms in hydroponically-grown flat-leaf parsley

This study uses supervised machine learning to combine 3D, multispectral, and thermal imaging data for improved detection of root rot in hydroponic parsley.

MaiZaic: a robust end-to-end pipeline for mosaicking freely flown aerial video of agricultural fields

MaiZaic improves mosaic accuracy for UAV agricultural imagery by using deep learning for homography estimation and error reduction.

The Global Wheat Full Semantic Organ Segmentation (GWFSS) dataset

A 1,096-image pixel-labelled wheat organ dataset plus 52,078 unlabelled field images from 11 institutions enables Segformer leaf and spike segmentation near 90% mIoU, while stems lag.

Automated and high throughput measurement of leaf stomatal traits in canola

SCAN measures canola stomatal traits with 89-94% accuracy using YOLOv8 and digital microscopy, and predicts traits in other species.

PhenoVision: A framework for automating and delivering research-ready plant phenology data from field images

PhenoVision automates plant phenology data extraction from field images, achieving high accuracy in detecting flowers and fruits, and expands global phenology data coverage.

FieldDino: High-throughput physio-morphological phenotyping of stomatal characteristics for plant breeding research

Handheld porometry, a 3D-printed leaf-clip microscope and YOLOv8 segmentation jointly phenotype stomatal conductance and anatomy on the same wheat leaf across 200 field genotypes.

Spatio-temporal modeling of host-pathogen interactions using level set method

This paper develops a spatio-temporal model for plant disease phenotyping using image processing and the level set method to track lesion progression and host-pathogen interactions.