Early yield prediction in different grapevine varieties using computer vision and machine learning

Abstract

Yield assessment is a highly relevant task for the wine industry. The goal of this work was to develop a new algorithm for early yield prediction in different grapevine varieties using computer vision and machine learning. Vines from six grapevine (Vitis vinifera L.) varieties were photographed using a mobile platform in a commercial vineyard at pea-size berry stage. A SegNet architecture was employed to detect the visible berries and canopy features. All features were used to train support vector regression (SVR) models for predicting number of actual berries and yield. Regarding the berries’ detection step, a F1-score average of 0.72 and coefficients of determination (R2) above 0.92 were achieved for all varieties between the number of estimated and the number of actual visible berries. The method yielded average values for root mean squared error (RMSE) of 195 berries, normalized RMSE (NRMSE) of 23.83% and R2 of 0.79 between the number of estimated and the number of actual berries per vine using the leave-one-out cross validation method. In terms of yield forecast, the correlation between the actual yield and its estimated value yielded R2 between 0.54 and 0.87 among different varieties and NRMSE between 16.47% and 39.17% while the global model (including all varieties) had a R2 equal to 0.83 and NRMSE of 29.77%. The number of actual berries and yield per vine can be predicted up to 60 days prior to harvest in several grapevine varieties using the new algorithm.

Computer Vision and Deep Learning for Precision Viticulture

“…Palacios et al [145] studied this modeling step in detail and reported (1) 86% recall and 62.2% precision for visible berry detection and (2) 26% Normalized RMSE, R² = 0.82, for actual berry counting with a Support Vector Regression model with features based on fruit area and occlusion. This work was extended to yield prediction [146]. Different methods for image-based yield estimation are detailed and compared in the next section.…”

“…Features related to occlusion and fruit area were used as yield predictors in the work of Palacios et al [145,146]. An error rate for 6 varieties of 29.77% NRMSE (in kg per vine, R 2 = 0.83) was reported with images taken 66 days before harvest.…”

Weakly and semi-supervised detection, segmentation and tracking of table grapes with limited and noisy data

“…More recently, a good number of deep learning based detection algorithms have been proposed. Palacios et al (2022) present an early yield prediction system based on berry counting using a SegNet segmentation network to extract features from the bunches and canopy images, such as the number of estimated visible berries, or the ratio of leaves in the detection bounding boxes. The statistics on the variability of the features of the 6 grapevine varieties stresses how cross-variety distribution shifts are significant for learning techniques.…”

Correlation of the Grapevine (Vitis vinifera L.) Leaf Chlorophyll Concentration with RGB Color Indices

“…To achieve these goals, remote sensing-based decision support systems (DSSs) are usually established to provide information for differentiated nutrient and water supply, plant protection, and harvest [1]. Among passive devices, RGB, thermal, multispectral, and hyperspectral cameras are the most widespread devices used to receive information about biomass and plant physiology or to predict yield and, moreover, to support microclimatic monitoring and terroir zoning [2][3][4][5][6]. Physical traits of the plants such as size, shape, and color are investigated most frequently on the whole canopy or individual leaves, shoots, and moreover on bunches, berries, or even on seeds or roots.…”