A high-throughput phenotyping system using machine vision to quantify severity of grapevine powdery mildew

Authors: BIERMAN, ANDREW - Rensselaer Polytechnic Institute; LAPLUMM, TIM - Rensselaer Polytechnic Institute; Cadle-Davidson, Lance; GADOURY, DAVID - Cornell University; MARTINEZ, DANIEL - Cornell University; SAPKOTA, SURYA - Cornell University; REA, MARK - Rensselaer Polytechnic Institute

Submitted to: Plant Phenomics
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 7/17/2019
Publication Date: 8/25/2019
Citation: Bierman, A., Laplumm, T., Cadle Davidson, L.E., Gadoury, D., Martinez, D., Sapkota, S., Rea, M. 2019. A high-throughput phenotyping system using machine vision to quantify severity of grapevine powdery mildew. Plant Phenomics. 1..

Interpretive Summary: The future of visual plant trait measurement uses high-throughput machine vision techniques. However, powdery mildews present specific challenges for such image-based analysis. We previously developed low-throughput microscopy approaches to measure resistance to grape powdery mildew for genetic analysis. Here, we developed nondestructive automated imaging methods followed by computer vision analysis to measure resistance to grape powdery mildew. The system images 135 to 270 samples per hour at high resolution. A neural network based on GoogLeNet then detects the presence of powdery mildew in approximately 800 subimages per sample, with an accuracy of 94.3%. The neural network was in agreement with human experts on the presence of powdery mildew for 89.3 to 91.7% of sub-images. This live-imaging approach enables a time course of infection and distinguishes susceptible, moderate, and resistance samples. With high throughput and accuracy, this phenotyping system could find application in numerous powdery mildew treatments. In addition, new neural networks could be developed for other plants, diseases, and traits.

Technical Abstract: Powdery mildews present specific challenges to phenotyping systems that are based on imaging. Having previously developed low-throughput, quantitative microscopy approaches for phenotyping resistance to Erysiphe necator on thousands of grape leaf disk samples for genetic analysis, here we developed automated imaging and analysis methods for E. necator severity on leaf disks. By pairing a 46 megapixel CMOS sensor camera, a long-working distance lens providing 3.5× magnification, X-Y sample positioning, and Z-axis focusing movement, the system captured 78% of the area of a 1-cm diameter leaf disk in 3 to 10 focus-stacked images within 13.5 to 26 seconds. Each image pixel represented 1.44 µm2 of the leaf disk. A convolutional neural network (CNN) based on GoogLeNet determined the presence or absence of E. necator hyphae in approximately 800 subimages per leaf disk as an assessment of severity, with a training validation accuracy of 94.3%. For an independent image set the CNN was in agreement with human experts for 89.3% to 91.7% of subimages. This live-imaging approach was non-destructive, and a repeated measures time-course of infection showed differentiation among susceptible, moderate, and resistant samples. Processing over one thousand samples per day with good accuracy, the system can assess host resistance, chemical or biological efficacy, or other phenotypic responses of grapevine to E. necator. In addition, new CNNs could be readily developed for phenotyping within diverse pathosystems or for diverse traits amenable to leaf disk assays.