A Non-Invasive Method Based on Computer Vision for Grapevine Cluster Compactness Assessment Using a Mobile Sensing...

Area of Science:

• Agricultural Engineering
• Computer Vision
• Viticulture

Background:

• Grapevine cluster compactness significantly impacts grape composition, disease susceptibility, and wine quality.
• Traditional cluster compactness assessment relies on subjective visual inspection, limiting practical application in the wine industry.
• Objective and efficient methods are needed for real-time cluster compactness evaluation in commercial vineyards.

Purpose of the Study:

• To develop and validate a novel, non-invasive method for assessing grapevine cluster compactness.
• To integrate computer vision and machine learning for automated compactness estimation using field-acquired RGB images.
• To provide a tool for the wine industry enabling efficient and objective cluster compactness assessment.

Main Methods:

• A mobile sensing platform captured on-the-go red, green, blue (RGB) images of grapevine clusters under artificial illumination.
• A semi-supervised image segmentation algorithm was employed for initial image processing, followed by automated cluster detection.
• Gaussian process regression modeled cluster compactness, with models calibrated and validated using expert ratings (OIV 204 standard).

Main Results:

• The developed computer vision method achieved a coefficient of determination (R²) of 0.68 and a root mean squared error (RMSE) of 0.96 on the test set.
• Leave-one-out cross-validation (LOOCV) demonstrated robust performance with R² of 0.70 and RMSE of 1.11.
• The image-based estimations showed strong correlation with average expert ratings for cluster compactness.

Conclusions:

• The developed computer vision and machine learning approach provides a reliable and non-invasive method for grapevine cluster compactness assessment.
• This technology is suitable for commercial application in the wine industry, utilizing on-the-go RGB image acquisition.
• The method offers potential for improved grape quality management and wine production efficiency.