Going Beyond RGB-Based Machine Learning Datasets for Digital Agricultural Applications

This project represents the next innovative leap by the TerraByte research group at UWinnipeg and the Enterprise Machine Intelligence and Learning Initiative (EMILI). The focus main thrust of this work is to generate data that will enable machine learning approaches to digital agricultural applications. The data generated from this project includes multispectral, hyperspectral, and 3D data. This data is fundamental to developing machine learning algorithms that require thousands to missions of examples in order to “learn” to perform specific tasks. The main applications that will be driving the data collection in this project are identifying weeds from prairie cash crops (called plant classification), recognizing specific plant traits and characteristics (called phenotyping) and identifying sickly plants (called disease detection). The benefits of this work are immense and include bringing impactful increases in agricultural production and global food security, which is especially important in a post-pandemic era.

Faculty Supervisor:

Christopher Bidinosti

Student:

Partner:

Enterprise Machine Intelligence and Learning Initiative

Discipline:

Computer science

Sector:

Agriculture

University:

University of Winnipeg

Program: