Skip to main content
ARS Home » Northeast Area » Beltsville, Maryland (BARC) » Beltsville Agricultural Research Center » Mycology and Nematology Genetic Diversity and Biology Laboratory » Research » Publications at this Location » Publication #431767
Research Project: Plant-associated Nematode Management and Systematics and USDA Nematode Collection Curation

Location: Mycology and Nematology Genetic Diversity and Biology Laboratory

Title: Beech leaf disease symptom detection using deep learning and computer vision tools

Author
item Waldo, Benjamin
item Reis Vieira, Paulo
item BORDEN, MATTHEW - Bartlett Tree Company
item Li, Shiguang

Submitted to: Journal of Nematology
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 3/19/2026
Publication Date: 4/26/2026
Citation: Waldo, B.D., Reis Vieira, P.C., Borden, M., Li, S. 2026. Beech leaf disease symptom detection using deep learning and computer vision tools. Journal of Nematology. 58(1):66-77. https://doi.org/10.2478/jofnem-2026-0011.
DOI: https://doi.org/10.2478/jofnem-2026-0011

Interpretive Summary: Beech leaf disease (BLD) is rapidly spreading across the eastern United States, where it can cause severe canopy decline and eventual tree death. Diagnosis currently relies on visually identifying the distinctive dark banding that forms between leaf veins, which is not possible at certain tree heights or at early stages. Although artificial intelligence (AI) has been increasingly used to support plant disease identification, no AI-based system has previously been developed for detecting BLD from images. Researchers at ARS trained a machine learning model capable of identifying BLD in real-world images with over 95% accuracy. This work provides an important step toward a more comprehensive BLD detection system and establishes a foundation for future image-based diagnostics of foliar nematode diseases. This will assist tree and forest health professionals in rapidly identifying the presence of the disease and enhance their ability to control the spread of this devastating tree disease.

Technical Abstract: Beech leaf disease (BLD) has rapidly emerged as a significant threat to forests across the eastern United States and Canada, and early detection is a major challenge. Current surveillance relies on visual identification of characteristic leaf banding, a method that can miss early infections. To address this limitation, we developed deep learning models capable of distinguishing between BLD symptomatic leaves and asymptomatic leaves. A primary dataset of symptomatic and asymptomatic leaves collected in Maryland (Dataset I) was used for model development and an independent set of images collected in North Carolina (Dataset II) provided real-world validation. In Dataset I model testing, EfficientNetV2-Small was the most accurate model (100%), followed by ResNet50 (99.32%) MobileNetV3-Large (97.95%), and InceptionV3 (94.88%). Independent testing on Dataset II also identified EfficientNetV2 Small as the most accurate model (96.55%). Grad-Cam visualizations confirmed that EfficientNetV2-Small focused on banded regions of BLD leaves that are characteristic of the disease. These findings demonstrate the potential of deep learning and computer vision approaches to support more efficient monitoring of BLD in forested landscapes.