US 12,333,424 B2
Determining cereal grain crop yield based on cereal grain trait value(s)
Zhiqiang Yuan, San Jose, CA (US); and Theodore Monyak, San Francisco, CA (US)
Assigned to Deere & Company, Moline, IL (US)
Filed by Deere & Company, Moline, IL (US)
Filed on Oct. 16, 2020, as Appl. No. 17/072,941.
Prior Publication US 2022/0121919 A1, Apr. 21, 2022
Int. Cl. G06N 3/08 (2023.01); G01N 33/00 (2006.01); G01N 33/10 (2006.01); G06V 20/10 (2022.01)
CPC G06N 3/08 (2013.01) [G01N 33/0098 (2013.01); G01N 33/10 (2013.01); G06V 20/188 (2022.01)] 14 Claims
OG exemplary drawing
 
1. A method implemented by one or more processors, the method comprising: for an image of at least a portion of a cereal grain plot in a sequence of images captured during a portion of a growing season less than an entire growing season, determining, by at least one processor circuit programmed by at least one instruction, a number of cereal grain spikelets in the image, wherein determining the number of cereal grain spikelets in the image includes: processing the image to determine a trait of a plant in the image of the cereal grain plot, wherein processing the image includes: using a spikelet detection model to identify one or more cereal grain spikelets in the image, the spikelet detection model to identify the one or more cereal grain spikelets in the image based on a training instance, the training instance to train the spikelet detection model using a training image portion and a ground truth image portion, the ground truth image portion to include an identification of a position of a cereal grain spikelet of a cereal grain plant shown in the training image portion; and generating a bitmap including a green region and a non-green region of the image, wherein the non-green region corresponds to a maturity of a plant of the cereal grain plot; and determining the number of cereal grain spikelets in the image based on the identified one or more cereal grain spikelets and the bitmap; determining a heading value for the image based on at least the number of cereal grain spikelets in the image, wherein determining the heading value for the image based on at least the number of cereal grain spikelets in the image includes: determining a change between the number of cereal grain spikelets in the image and the number of cereal grain spikelets in one or more previous images in the sequence; comparing the change in the number of cereal grain spikelets with a heading growth chart of the cereal grain plot for one or more previous growing seasons, where the heading growth chart is based on processing of a previous sequence of images capturing a previous growing season for the cereal grain plot; and determining the heading value for the image based on the comparison; and determining, by one or more of the at least one processor circuit, a predicted yield for the cereal grain plot based on at least one or more of the determined heading values.