Development of digital technologies for wild blueberry cropping system to lower production costs and increase berry quality

Artificial intelligence coupled with machine vision agrochemical sprayers can replace traditional uniform applications. Novel advancements utilizing high resolution images with deep learning techniques are required to develop new algorithms for advanced real-time automated classification. Fields will be surveyed, and a digital library database of images will be acquired for the major target weeds that are spatially variable throughout the fields. Following evaluation of the deep-learning program in the lab and static field environment, the system will be integrated onto a commercial uniform sprayer modified to operate with individual nozzle control for precise weed target detection. An economic analysis will compare the costs of uniform spraying methods to the new spot-application techniques. Deep learning models will also be developed for detecting berry growth and crop diseases. Software will be designed and deployed for use on mobile phones that will process field images to identify weeds, diseases, and plant development in real-time.

Faculty Supervisor:

Travis Esau

Student:

Patrick Hennessy

Partner:

Wild Blueberry Producers Association of Nova Scotia

Discipline:

Engineering

Sector:

Agriculture

University:

Dalhousie University