A novel dataset and deep learning object detection benchmark for grapevine pest surveillance

Riesling is the iconic grape variety of Germany and accounts for 23% of the German viticulture acreage, which comprises 45% of the worldwide Riesling plantings. Riesling wines offer a wide array of styles from crisp sparkling wines to highly concentrated and sweet Trockenbeerenauslese or Icewines. However, its thin berry skin makes Riesling more vulnerable to detrimental environmental threats than other white wine varieties.  

Botrytis cinerea, an Ascomycota pathogen with a broad host range, infects over 1200 plant species. Grapes infected by this pathogen, which subsequently develop a noble rot, remain in the vineyard for an extended period, thus being exposed to a diverse array of physical, chemical and biological factors, which give rise to a complex microbial community.

Agriculture is one of the largest contributors to environmental pollution and causing significant impacts on human health, ecosystems, and resource availability.

Cabernet Sauvignon is the top red wine cultivar in CA, however, the hot climate in Fresno is not ideal for Cabernet Sauvignon, particularly for berry color development. Fruit-zone leaf removal and irrigation were studied previously to have the significant effect on grape yield performance and berry quality. But the timing of leaf removal and the timing of irrigation are still inconclusive. Also, mechanical fruit-zone leaf removal is relatively new in CA. Our study aims to identify the interactive effect of mechanical fruit-zone leaf removal and irrigation on Cabernet Sauvignon’s yield performance and fruit quality and find the ideal timing of leaf removal and irrigation to maximize the berry color while maintaining the sustainable yield level.

A large varietal collection including over 1700 varieties was maintained in Conegliano, ITA, since the 1950s. Phenological data on a subset of 400 grape varieties including wine grapes, table grapes, and raisins were acquired at bud break, flowering, veraison, and ripening since 1964. Despite the efforts in maintaining and acquiring data over such an extensive collection, the data set has varying degrees of missing cases depending on the variety and the year. This is ubiquitous in phenology datasets with significant size and length. In this work, we evaluated four state-of-the-art methods to estimate missing values in this phenological series: k-Nearest Neighbour (kNN), Multivariate Imputation by Chained Equations (mice), MissForest, and Bidirectional Recurrent Imputation for Time Series (BRITS). For each phenological stage, we evaluated the performance of the methods in two ways. 1) On the full dataset, we randomly hold-out 10% of the true values for use as a test set and repeated the process 1000 times (Monte Carlo cross-validation). 2) On a reduced and almost complete subset of varieties, we varied the percentage of missing values from 10% to 70% by random deletion. In all cases, we evaluated the performance on the original values using normalized root mean squared error. For the full dataset we also obtained performance statistics by variety and by year. MissForest provided average errors of 17% (3 days) at budbreak, 14% (4 days) at flowering, 14.5% (7 days) at veraison, and 17% (3 days) at maturity. We completed the imputations of the Conegliano dataset, one of the world’s most extensive and varied phenological time series and a steppingstone for future climate change studies in grapes. The dataset is now ready for further analysis, and a rigorous evaluation of imputation errors is included.