Drone-based phenotyping platform for white mold resistance in dry beans
Aerial imaging and AI-driven analytics platform using convolutional neural networks and crop growth models to detect and evaluate white mold resistance in dry bean breeding lines. Enables automated, scalable disease phenotyping to accelerate genomic selection and reduce yield losses.
Overview
White mold, caused by Sclerotinia sclerotiorum, is a major yield-limiting disease in dry bean (Phaseolus vulgaris) and other crops. Traditional visual disease scoring is labor-intensive, subjective, and difficult to scale across large breeding populations. This solution combines unmanned aerial system (UAS) imagery with convolutional neural network (CNN) and crop growth model (CGM) analytics to deliver objective, high-throughput phenotyping of white mold resistance. By correlating temporal multispectral and RGB imagery with disease severity and grain yield data, the platform supports breeders in identifying resistant genotypes more efficiently and improving genomic selection pipelines.
Technical specifications
- Aerial data acquisition: Multispectral and RGB sensor flights conducted across key reproductive growth stages (R1–R8), with weekly flights after R1 for temporal monitoring
- CNN-based disease detection: Deep learning models trained on UAS imagery to automatically classify disease severity and identify resistant plots
- Crop growth modeling: Integration of CGM with image-derived agronomic traits such as canopy coverage, plant height, and maturity to characterize disease progression
- Phenotypic trait extraction: Assessment of agronomic traits from RGB imagery combined with multispectral data for plant disease identification
- Genomic selection support: Genotypic growth curves correlated with severity scale measurements to enhance selection efficiency in breeding programs
- Experimental design: Validated on 176 advanced breeding lines arranged in an augmented incomplete block design with two replications
Technology readiness level
The platform has been validated in a 2021 field season trial across 176 advanced breeding lines at Michigan State University. Future validation is planned across diverse breeding panels during the 2021 and 2022 crop seasons, with continued refinement of CNN training datasets and CGM integration. The approach is being developed in coordination with MSU advanced breeding lines and the National Sclerotinia Initiative, positioning it for broader adoption in public and private breeding programs. The team is seeking partnerships for pipeline development, equipment acquisition, and key personnel funding to advance deployment.
About Michigan State University
Michigan State University is a major public land‑grant research university with a comprehensive academic portfolio and a large research enterprise. Industry partners engage through an on‑campus U.S. Department of Energy national user facility and shared core laboratories with user access. The university provides a chemical process scale‑up pilot plant on Michigan’s lakeshore, a research and technology park, and a Grand Rapids health innovation campus linking researchers with clinical partners. A statewide extension network supports field deployment and workforce training across Michigan’s manufacturing corridor. Research is backed by competitive federal funding from NSF, NIH, DOE, USDA, and DoD, while dedicated tech transfer and corporate engagement teams—supported by an affiliated research foundation—accelerate IP, licensing, startups, and sponsored research.
delivered to your inbox.