Research Highlights

Research Highlights
Bones, Beans and Building First-of-its-Kind Quality-Assessment Machine

New technology is being developed to analyze the shape, color and texture of soybeans. Source: Pixabay, photographer Ebreh Snbnhehko

By Randy LaBauve, LSU AgCenter

Reprinted from LSU AgCenter website, Aug. 29, 2023

What do bones and soybeans have in common? They both have defining traits that can be analyzed through imaging, and just as Louisiana State University AgCenter engineer Kevin Hoffseth has digitally imaged and analyzed bone quality for many years, he’s relied on that specialized expertise to invent a technological system for improved evaluation of soybean quality.

“What I do in both of these areas is connected by trying to say something about how good something is,” said Hoffseth. “We’re putting together a machine to distinguish and assess soybeans in a way that hasn’t been done before, as far as I know.”

Soybean producers have been concerned with undue variation of inspection and grading — whether the process was accurately reflecting the quality and grade of their beans. Hoffseth, in 2020, began research targeting an automated image-based system to develop more consistent, accurate assessments of quality, with support from the Louisiana Soybean and Grain Research and Promotion Board.

“The warehouse inspector is put under the continual stress of analyzing large numbers of truckloads at harvest time,” said Hoffseth. “Inspection is a manual process where a person judges visual factors that indicate quality, and that rating affects the money that farmers can make.”

This process graphic shows how the quality-assessment machine assesses soybean quality. Source: Kevin Hoffseth, Louisiana State University AgCenter

Inspections assess bean quality based upon color, texture, shape and any notable damage, such as visible mold or shriveling. Inspectors use USDA standards to rate a batch of soybeans as one of five different grades.

The first couple years of the project were spent researching and developing image processing algorithms and programming implementation. Hoffseth and his graduate students also worked on acquiring and building hardware, such as specialized camera stations, to build visual-based setups for quality assessment that should assist the grading process.

“The programming needs to be set up to see the same thing each time and that saves a lot of work, saves a lot of problems and makes it the quickest,” said Hoffseth.

Most recently, Hoffseth’s team has been creating prototype machinery to, among other things, address:

  • Analyzing texture.
  • Developing mechanisms to move grain past the cameras.
  • Establishing greater speeds for imaging.
  • Establishing ideal resolution of images.
  • Determining the best image compression for saving useful information.
  • Controlling lighting for consistent results.
  • Creating an optimized image storage system.

“We’ve been working to develop new and improved algorithmic approaches for image processing analysis methods to look at these soybeans and give quantitative values,” said Hoffseth. “We’re looking forward to getting these prototypes finished so that we can do some initial experiments at farms.”

Hoffseth and the LSU AgCenter Office of Intellectual Property have applied for a provisional patent for the technologies they’ve already developed. A crucial part of the project’s future success, however, will be the creation of a large database of soybean images used alongside big-data applications, said Hoffseth.

“That becomes super valuable because there’s this immediate need of building a database of our locally grown varieties in Louisiana and understanding the trends,” said Hoffseth. When they have these stored images, they can run analyses on them, keep them for future reference and then run them on future applications.

Soybean producers have also approached Hoffseth about assessing oil and protein content in their beans. That could be done with various types of infrared technologies.

“That’s a wonderful thing that can come as an outgrowth of this work,” said Hoffseth. “As of now, inspecting oil is not part of the inspection and grading process, but farmers, when their beans are damaged, want to know whether they could still have value because of good oil content.”

Hoffseth said these technologies are not meant to replace the inspector, but to offer tools for helping quickly come up with consistent and accurate results. Louisiana harvested nearly 1 million acres of soybeans in 2020, so this new customized technology could be very useful to many farmers in the state, and even to farmers nationwide. The invention also holds potential for use with other similarly sized grain crops, according to Hoffseth.

“If we can quantify and take out a lot of variability and subjectivity in how those soybeans are graded,” Hoffseth said, “it can be a huge factor in making everyone’s life easier.”

Additional Resources

Digital Image Processing Addresses Consistency of Soybean Grading – SRIN article

Published: Apr 1, 2024

The materials on SRIN were funded with checkoff dollars from United Soybean Board and the North Central Soybean Research Program. To find checkoff funded research related to this research highlight or to see other checkoff research projects, please visit the National Soybean Checkoff Research Database.