University Administration Building
701 S. Nedderman Drive, Ste 421
Arlington, TX 76019-0116
Scene
West Texas Cotton Fields
Cotton is Texas’ top cash crop, with an economic impact of $5.2 billion, much of it from West Texas. However, rising heat and drought conditions are challening farmers. To help, agronomists and cotton breeders are developing heat-tolerant cultivars—with UTA researchers playing a key role.
“This process takes many years,” says William Beksi, assistant professor of computer science and engineering. “They have to grow the cultivars in a small portion of a field over many seasons to see if they have the phenotypes or traits they’re looking for, and then ultimately decide to breed them at a large scale.”
Making that decision comes down to phenotyping data, which is typically gathered manually by workers who count the number of cotton bolls produced by the cultivar, then extrapolate that number out to the size of a field.
“It’s error-prone, with humans out there in the hot, dusty environment doing the count manually. It’s boring, and mistakes happen,” Dr. Beksi says. “What we want to try to do is to see if we can help them automate that process.”
Beksi and his team, including students, tested an unmanned ground system in fields maintained by Texas A&M and Texas Tech. They developed computer vision algorithms to detect and count cotton bolls from video data. The work, says Beksi, is both professionally and personally gratifying.
“It’s extremely technically challenging, so there are a lot of really great technical problems to solve here,” he says. “But to do work in agriculture especially is really motivating because you know the world can really benefit from that research.”
1. FIELD LEADER
William Beksi, project leader, follows the unmanned ground vehicle (UGV) as it navigates rows of cotton crops. The UGV carries stereo cameras that record field data for further analysis.
2. MACHINE VISION
The unmanned ground vehicle is equipped with dual stereo cameras that capture symmetric views of the crop canopy for tracking and counting cotton bolls.
3. DATASETS
UTA researchers are recording sets of cotton crop video sequences for training and testing multi-object tracking methods.
4. THE COTTON BOLL
In the field, researchers are helping plant scientists determine which cultivars produce the largest, most abundant, and heartiest crops.