Click on “Download PDF” for the PDF version or on the title for the HTML version.


If you are not an ASABE member or if your employer has not arranged for access to the full-text, Click here for options.

Prediction of Citrus Yield With AI Using Ground-Based Fruit Detection and UAV Imagery

Published by the American Society of Agricultural and Biological Engineers, St. Joseph, Michigan www.asabe.org

Citation:  2021 ASABE Annual International Virtual Meeting  2100493.(doi:10.13031/aim.202100493)
Authors:   Vinay Vijayakumar, Lucas Costa, Yiannis Ampatzidis
Keywords:   Click here to enter keywords and key phrases, separated by commas, with a period at the end

Abstract. Yield prediction of citrus provides critical information before harvest to growers and stakeholders to predict the resources required for workers, storage, and transportation of the harvest. Yield estimation using machine vision techniques has been applied in the past using fruit count from multiple tree images. Studies conducted on other fruits such as mangoes used unmanned aerial vehicles (UAVs) to collect tree structure parameters data (e.g., tree height, canopy area, and volume) to generate a fruit load index; however, machine vision-based fruit count was not included in these predictive production models. In this study, a combination of UAV aerial data and fruits counted from ground images by a deep learning algorithm was used to generate an ensemble model for yield prediction. The UAV captured multispectral images, which provide the Blue, Green, Red, Near Infrared, and Red Edge band spectral data. Two images per tree were captured from the ground using an RGB camera (one from each side) and were used for fruit count using an object detection algorithm, YOLO V4. Harvest data was collected manually per tree (fruit count and weight). Four Machine Learning (ML) algorithms - Gradient Boosting Regression (GBR), Random Forest Regression (RFR), Linear Regression (LR), and Partial Least Squares Regression (PLSR) were used to generate the model. The combined UAV-ground yield prediction model generated was compared with the “UAV only” model. The yield prediction model based on the sensor fusion of image fruit count and UAV data gave a 24% reduction in error.

Keywords. Citrus, Machine Learning, Regression, Sensor Fusion, UAV, Yield prediction.

(Download PDF)    (Export to EndNotes)