A biomedical engineer accustomed to studying the composition of human bone is turning his eye toward determining the quality of soybeans.

Kevin Hoffseth, an assistant professor in the LSU AgCenter Department of Biological and Agricultural Engineering, specializes in the analysis of the mechanics and deformation of biological materials.

This past year, Hoffseth was awarded a grant from the Louisiana Soybean and Grain Promotion Board to develop a high-tech grading system for soybeans, and Hoffseth has been working with input from soybean producers and the LSU AgCenter to learn how to improve the current system.

“They say there is an issue with how they do the grading when producers go to sell their crop,” he said. “That can be very frustrating. I’m not a producer, but I can empathize with that.”

In his discussions with soybean farmers, Hoffseth heard that one truckload of beans could get three different grades from three different inspectors. Hoffseth thinks he can apply image processing and analysis techniques from the engineering world to assist inspectors.

“You import the images into a computer, and you have a set of algorithms that look at each image the same way each time,” Hoffseth said. “The thinking is that we hope we can develop some methods and some hardware later on in the next year or two to help the inspectors have some more consistency in the grading.”

Hoffseth does not think that computers can replace human inspectors.

“There are reasons for people, and people are great,” Hoffseth said. “We are trying to improve the repeatability and cut down on inconsistencies and try to remove different sorts of sampling conditions and try to remove as much gray area as possible.”

Soybeans are divided into four numerical grades, No. 1 through No. 4, according to the U.S. Department of Agriculture Federal Grain Inspection Service standards. The soybeans are graded on weight, the percentage of damaged kernels, color and other specifications. A fifth grade, called sample grade, includes soybeans of lower quality.

The amount of foreign material or off-color beans present in the load may also affect the grade. Searching for different shapes and colors to identify foreign material is one place computers and cameras excel, Hoffseth said.

“With cameras and computers, they can do it faster than people,” he said. “People still need to oversee the process. Proper computer algorithms can match shapes and identify objects very fast. Let’s use that horsepower.”

Hoffseth has had little time for full-scale research and development. He and his co-principal investigator, Dorin Boldor, received funding for the grant in April. They faced restrictions into summer because of the COVID-19 pandemic.

While his research has just begun, Hoffseth doesn’t foresee robots taking over for humans while grading soybeans.

“Maybe we can design some sort of hardware, camera and lighting that you just put the beans in and it will automatically take an image for quality,” Hoffseth said.

Such a system would be designed to cut down on human error.

“It’s not left up to a tired inspector out in the sun or in a barn or in a small office somewhere. The cameras don’t care. They just evaluate everything the same way every time. We really want to help the people make it better and easier,” Hoffseth said.

This story is featured in the Louisiana Soybean and Grain Research and Promotion Board 2020 Report.

Kevin Hoffseth, an assistant professor in the LSU AgCenter Department of Biological and Agricultural Engineering, specializes in the analysis of the mechanics and deformation of biological materials.

According to U.S. Department of Agriculture Federal Grain Inspection Service standards, soybeans are divided into two colors, yellow and other. Soybeans are also divided into four numerical grades, No. 1 through No. 4. Photo courtesy of the USDA.