Final Report: June 30, 2021
Manoj Karkee, Center for Precision and Automated Agricultural Systems, Washington State University
Project Summary: Crop yield estimation is an important managerial tool for vineyard manager (Read and Gamet, 2007) and plays a crucial role in planning pre/post-harvest operations to achieve desired crop-load and to improve efficiency and reduce cost of various field operations. Numerous approaches have been reported in literature such as the one from Robotics Institute, Carnegie Mellon University for automated yield estimation in wine grapes. However, growers have adoption concerns as the technology is expensive, needs high technical expertise and yet has not proved to be sufficiently accurate.
Our approach is a simple and low-cost yet practical design of a sensing system for crop and yield estimation in wine grapes. A Software Application (App) was developed that used the sensors (e.g. cameras) of a mobile device (e.g., a smartphone) to acquire images in vineyards. These images were uploaded to a distant server where they were processed in near-real time for detection, counting, and sizing of berries in the cluster. The App then was used to scan and count berries and clusters in several sample vines in a plot. This tool, in the current form, can be used to minimize labor use in crop and yield estimation by automating cluster and berry counting, and berry sizing operation in sample clusters. In the future (with the new funding our team recently received from the commission), the tool will be further developed to be used as an input to geostatistical and crop growth models for estimating overall yield in vineyards. When successful, this technology could be available commercially to growers for small cost as it runs on users’ existing hardware (e.g., compatible smartphones). Because smartphones and tablets are ubiquitous and pre-equipped with necessary sensors such as cameras and GPS, an App-based, low-cost approach has a great potential for in-hand and near-real time crop-load estimation.
Read more by downloading the full report above.