Estimating Apple Crop Yield using Images

As a high-value crop, apples are intensively managed with much of this management being associated with yield estimates. Currently, estimates are done manually by experience farmers and by Scotian Gold experts. Unfortunately, this results in costly estimates that vary in their quality from farm to farm and between apple varieties. A system that can provide consistent and accurate estimates using low cost digital photography and cloud-based artificial intelligence can provide an opportunity to improve production and marketing for the industry. The objectives of this project are: (1) to develop a method of capturing images of an apple variety on the tree in the mid-August time frame; (2) to develop a computational vision and machine learning prototype for estimating the size and number of apples on a tree using the images captured; and (3) to use these estimates to predict the harvest yield (bushels per acres) adjusting for forecasted weather.

Faculty Supervisor:

Daniel Silver

Student:

Niroj Shrestha

Partner:

Scotian Gold Cooperative

Discipline:

Computer science

Sector:

Professional, scientific and technical services

University:

Acadia University