Skip to main content Skip to navigation

In-tree apple crop-load estimation with smartphones

Sponsor Agency: WSU Office of Commercialization
WSU Investigators: Karkee, M.

 

Crop-load estimation is an important management tool for apple growers. It is crucial for the efficient management of pre/post-harvest operations such as labor and equipment requirements for harvesting and transporting fruit from the orchard to packing house. Various research efforts around the world have developed machine vision systems for apple detection and counting, but the sensors (e.g. industrial cameras and GPS units) and platforms/machines used for crop-load estimation research have been complex and expensive while also requiring a skilled and dedicated operator to collect field data. These limitations of the sensing system and platforms have created a bottleneck for commercial adoption of the crop-load estimation technology despite a good level of fruit counting and sizing accuracy. In this project, we are developing a simple and low-cost yet practical approach for apple crop-load estimation. A Software Application (App) was developed that used the sensors (cameras and GPS) of a mobile device (e.g. a smartphone) to acquire images in apple orchards. These images are processed in near-real time for detection, counting, and sizing of apples. The App is then be used to scan and count fruit in a number of sample trees in a plot, which will be used as an input to a geostatistical model developed for estimating crop-load in orchards. A preliminary work conducted in a commercial orchard with the newly developed App showed an accuracy of 98% in estimating number of apples in a fruiting wall orchard.