On March 3, 2022, John Deere introduced See and Spray Ultimate, a factory-installed system for its self-propelled sprayers that uses computer vision and machine learning to tell crops from weeds and spray non-residual herbicide only where a weed is detected. The vision technology came from Blue River Technology, the machine-learning startup Deere acquired in 2017.
Boom-mounted cameras scan the ground as the sprayer moves at up to 15 miles per hour, covering more than 2,100 square feet of crop per second. Onboard processors decide whether each plant is a crop or a weed and send commands to individual nozzles, which deliver a precise dose of herbicide only on the recognized weeds. The alternative - broadcast spraying - covers the entire field whether or not weeds are present.
Deere reported large reductions in herbicide use. In a September 2024 update covering the 2024 growing season, the company said farmers running See and Spray across more than one million acres saved about 8 million gallons of herbicide mix, averaging 59 percent savings on corn, soybean and cotton fields. Deere cited an Iowa State University study that found 76 percent product savings across all test fields.
For row-crop farming the system reframed weed control as a perception problem: instead of treating a field as a uniform surface, the machine makes a crop-or-weed decision plant by plant, in real time, at field speed. It also produces a weed-pressure map in the John Deere Operations Center that farmers can compare against yield maps to plan future weed control.