Morphological image analysis for determining bunch grape characteristics: A case study on bunch weight in Cabernet-Sauvignon

Abstract

Morphological image analysis is a powerful technique used in various fields, including agriculture, to quantitatively assess the physical characteristics of objects. In viticulture, the accurate assessment of grapevine characteristics is essential for optimizing crop management and improving the quality of wine production. Among these characteristics, bunch weight is a critical factor influencing vine health, yield potential, and the quality of grapes harvested. Accurate vineyard yield estimation is crucial for the wine industry as it enables optimization in harvest planning, winery management, and marketing strategies (Victorino et al., 2022). However, the significant spatial and temporal variability within vineyards complicates precise predictions of grape bunch weight (Bramley et al., 2011). Conventional methods, such as manual grape bunch sampling, are destructive, labour-intensive, and prone to significant errors that can exceed 30%, depending on the sampling technique used and vineyard heterogeneity (Dunn & Martin, 2008). To overcome these limitations, sensor-based technologies, particularly image analysis, have shown great potential in addressing these challenges. These tools enable the inspection of a large number of grape bunches within a short time, reducing reliance on extrapolations and errors associated with variability (Liu & Zeng, 2020). Non-contact measurements based on two-dimensional image processing have been proven to be useful in the detection of several key agricultural traits, especially in single fruits with regular and uncomplicated shapes, such as apples and apricots (e.g., Khojastehnazhand et al., 2019; Wu et al., 2019). Other studies also present the potential of these techniques for more complicated fruit shapes, such as grape bunches, where the shape and size are highly dependent of the cultivar, viticulture practices and edaphoclimatic conditions. Diago et al., (2014) demonstrated that features such as the projected area of the bunch, the number of visible berries, and the perimeter are key predictors of bunch weight in two-dimensional analyses, achieving significant correlations across various cultivars. Moreover, the use of two-dimensional imaging has become an effective tool for the automatic segmentation of grape bunches and the counting of visible berries (Aquino et al., 2018; Milella et al., 2018). Advanced methods, such as algorithms based on convolutional neural networks, have significantly improved segmentation and counting under field conditions, bringing these technologies closer to practical applications in commercial vineyards (Liu & Zeng, 2020). However, the dependence of these correlations on the cultivar remains a challenge, as differences in bunch architecture and environmental conditions significantly affect the accuracy of proposed models (Tello et al., 2015; Victorino et al., 2022). Despite these advances, several open questions persist regarding image-based weight estimation. These include berry occlusion, variability in image capture conditions, and cultivar dependence, all which limit model generalization (Diago et al., 2014; Victorino et al., 2022).