Developing accurate estimates of crop yield and canopy size is a challenging, critical task for grapegrowers. Regardless of how you develop your estimates, crop weight and canopy estimates impact a variety of business decisions, including how to optimize fruit quality and manage the logistics at harvest. With accurate estimates, you can make more informed decisions.
From traditional techniques to computer vision
Traditionally, winemakers estimate crop weight by taking samples of the vineyard to calculate average number of clusters per vine, number of berries per cluster and average berry weight. The three are then multiplied to come up with a crop weight estimate. Traditional techniques like this can be destructive to the vine, labor intensive and subject to sampling issues.
Estimation techniques have become more advanced and less labor intensive. They range from satellite imagery to NDVI sensor-based estimates. Fruition Sciences wants to share with you an interesting study done by researchers at Cornell University. The research is part of a public private partnership funded by the USDA’s Specialty Crop Research Initiative (SCRI). The new technique employs laser scanners and computer vision to estimate canopy size and crop weight.
How the technique works
Canopy size measurement
Two laser scanners were mounted on a vineyard utility vehicle to capture 2D scans of vine rows at the rate of 75 scans/second. The researchers used the scans to project a 3D visualization of the whole vineyard. Using this 3D visualization, the volume each vine occupies can be estimated and used as a proxy for total leaf area, a key measurement of vine yield potential. The researchers found this method is capable of explaining 65% of pruning weight variation.
The researchers use a commercially available camera mounted on a vineyard utility vehicle. The vehicle ran down every row to capture images of the fruit zone every half second. Unlike the traditional sampling method, this new method captures the whole vineyard, thereby eliminating possible sampling issues. Then the researchers used computer vision algorithms to detect berry-like shapes and group those that belong to the same cluster in order to derive an estimate of berry number per vine. This method can explain 74% of yield variation.
Research partnerships like the SCRI are hugely beneficial to the winemaking industry. Wineries benefit from expertise of the academic community, while researchers benefit directly from working on problems that matter to winemakers. As a company started in the university research lab, Fruition Sciences strives to connect the research and industry communities around latest technologies in viticulture through this blog and at our Vintage Report conferences.
Fortunately many of these technologies have become available commercially to grape growers. Since 2015, Fruition Sciences has offered grape growers the ability to map vineyard biomass to optimize pruning and fertilizing decisions with our product Physiocap.