University of Winnipeg dives into agriculture research

Collaboration to develop expertise in high-tech precision ag technology

A greenhouse at the University of Winnipeg is growing plants for physics and computer science researchers working on machine-learning problems in precision agriculture.

A very urban university is starting to sink roots deep into Manitoba’s agriculture sector.

The University of Winnipeg is embarking on a collaboration with Enterprise Machine Intelligence and Learning Initiative (or EMILI as it’s known) to contribute to taking the agriculture industry high tech.

Ray Bouchard, EMILI chair and president and CEO of Enns Brothers, says the undertaking will build on strengths amongst the partner organizations.

“Our primary focus is working with our partners along with provincial and federal governments to secure funding and drive digital agriculture within the province of Manitoba,” Bouchard said. “Our work revolves around new technologies, intelligent technologies and finally the skills and talent component. The data research at the U of W had us very interested and that’s why we thought the U of W project would be a great jumping-off point for EMILI.”

Earlier this spring the endeavour got a big boost in the form of funds from the federal government. Navdeep Bains, minister of innovation, science and economic development at that time announced a total of $3,451,167 in funding for two projects to support innovation, skills development, and growth in the digital agriculture industry.

Prof. Christopher Henry (l to r), Minister of International Trade Diversification Jim Carr, University of Winnipeg president Annette Trimbee, Minister of Innovation, Science and Economic Development Navdeep Bains, Prof. Christopher Bidinosti at the announcement of precision agriculture research funding at the University of Winnipeg. photo: University of Winnipeg

Leading the research from the University of Winnipeg are physics professor, Christopher Bidinosti and applied computer science professor, Christopher Henry. Their research team includes experts from University of Winnipeg, Red River College, the University of Saskatchewan, Northstar Robotics, Sightline Innovation, the Canola Council of Canada, and Manitoba Pulse & Soybean Growers.

When asked why the U of W was interested in agricultural research despite lacking a strong connection to the industry, Bidinosti pointed out that the problem of data collection, digitalization, and automation is a global issue and can be applied to the agriculture industry among other sectors.

“The point is, this isn’t really an agricultural problem,” he said. “It’s a data problem, a machine learning problem, a computational problem, and it’s a sensors problem. We have skills in all of those areas as individual researchers, so it was actually very easy and efficient for use to focus on this looming problem.”

Bringing those expertise into the sector will give the U of W researchers and their students new problems to grapple with using their skills.

“We were in a position with all of our expertise in sensors, computing and data where it made a lot of sense, and we decided to jump on it,” he said.

Bidinosti says he hopes to show more students the potential of working in or with the agriculture industry, and in turn, show agricultural employers what type of skills they might look for and expect from new graduates.

So what will they be researching?

A few weeks ago the Manitoba Co-operator saw a prototype of a plant identification project. The goal of this project is to be able to look at examples of weed species and correctly identify the plant. The prototype uses cameras and identification software with thousands of previously collected images to identify the plant and weed species that we’d see here in Manitoba.

The problem?

A lot of plants look alike, even to the most experienced agronomists. Making sure the software will correctly identify a plant requires the researchers to take thousands of pictures of each different plant species at all of the growing stages.

The pictures of the plants at different stages go into a database for the software to reference when identifying a new plant. Their research team is currently in its most important phase, collecting and cataloguing the plant images.

The team hopes to have the collection done as soon as it can, noting that it will be an extensive process nonetheless. As more varieties and different crops appear in rotations, and new weeds crop up, having an accurate set of images for the ID software to work with is a crucial first step.

About the author


Spencer Myers is a former reporter with Glacier FarmMedia and a graduate of the University of Manitoba’s Agriculture Diploma Program and Red River College’s Creative Communications program.



Stories from our other publications