Yield estimation is an essential preharvest practice among most large-scale farming companies, since it enables the predetermination of essential logistics to be allocated (i.e., transportation means, supplies, labor force, among others). An overestimation may thus incur further costs, whereas an underestimation entails potential crop waste. More interestingly, an accurate yield estimation enables stakeholders to better place themselves in the market. Yet, computer-aided precision farming is set to play a pivotal role in this respect. Kiwifruit represents a major produce in several countries (e.g., Italy, China, New and Zealand). However, up to date, the relevant literature remains short of a complete as well as automatic system for kiwifruit yield estimation. In this paper, we present a fully automatic and noninvasive computer vision system for kiwifruit yield estimation across a given orchard. It consists mainly of an optical sensor mounted on a minitractor that surveys the orchard of interest at a low pace. Afterwards, the acquired images are fed to a pipeline that incorporates image preprocessing, stitching, and fruit counting stages and outputs an estimated fruit count and yield estimation. Experimental results conducted on two large kiwifruit orchards confirm a high plausibility (i.e., errors of 6% and 15%) of the proposed system. The proposed yield estimation solution has been in commercial use for about 2 years. With respect to the traditional manual yield estimation carried out by kiwifruit companies, it was demonstrated to save a significant amount of time and cut down on estimation errors, especially when speaking of large-scale farming.

This study was conducted to evaluate the potential of low-cost hyperspectral sensors for the early detection of Flavescence dorée (FD) from asymptomatic samples prior to symptom development. In total, 180 leaf spectra from 60 randomly selected plants (three leaves per plant) were collected by using two portable mini-spectrometers (Hamamatsu: 340–850 nm and NIRScan: 900–1700 nm) at five vegetative growth stages in a vineyard with grape variety Garganega. High differences in the Hamamatsu spectra of the two groups were found in the VIS-NIR (visible–near infrared) spectral region while very small differences were observed in the NIRScan spectra. We analyzed the spectral data of two sensors by using all bands, features reduced by an ensemble method, and by genetic algorithms (GA) to discriminate the asymptomatic healthy (FD negative) and diseased (FD positive) leaves using five different classifiers. Overall, high classification accuracies were found in case of the Hamamatsu sensor compared to the NIRScan sensor. The feature selection techniques performed better compared to all bands, and the highest classification accuracy of 96% was achieved when GA features of the Hamamatsu sensor were used with the logistic regression (LR) classifier on test samples. A slightly low accuracy of 85% was achieved when the features (selected by the ensemble method) of the Hamamatsu sensor were used with the support vector machine (SVM) classifier by using leave-one-out (LOO) cross-validation on the whole dataset. Results demonstrated that employing a feature selection technique can provide a valid tool for determining the optimal bands that can be used to identify FD disease in the vineyard. However, further validation studies are required, as this study was conducted using a small dataset and from the single grapevine variety.

Precision agriculture represents a promising technological trend in which governments and local authorities are increasingly investing. In particular, optimising the use of pesticides and having localised models of plant disease are the most important goals for the farmers of the future. The Trentino province in Italy is known as a strong national producer of apples. Apple production has to face many issues, however, among which is apple scab. This disease depends mainly on leaf wetness data typically acquired by fixed sensors. Based on the exploitation of artificial neural networks, this work aims to spatially extend the measurements of such sensors across uncovered areas (areas deprived of sensors). Achieved results have been validated comparing the apple scab risk of the same zone using either real leaf wetness data and estimated data. Thanks to the proposed method, it is possible to get the most relevant parameter of apple scab risk in places where no leaf wetness sensor is available. Moreover, our method permits having a specific risk evaluation of apple scab infection for each orchard, leading to an optimization of the use of chemical pesticides.