Remote-sensing-based detection of soybean aphid induced stress in soybean

Patent metadata

Field	Value
Publication number	US-11003908-B2
Application number	US-201916519233-A
Country	US
Kind code	B2
Filing date	Jul 23, 2019
Priority date	Jul 23, 2019
Publication date	May 11, 2021
Grant date	May 11, 2021

Abstract

A method of determining whether to treat soybeans for soybean aphids, the method includes collecting at least one image of a soybean canopy using one or more remote sensing instruments and processing the image into spectral reflectance data and selecting from the spectral reflectance data optimal spectral wavelength bands. The selected reflectance data is classified into one of a plurality of classification groupings using a machine learned classification model. To treat or not treat the soybean canopy for aphids is determined based on the classification of the reflectance data into one of the class groupings.

First claim

What is claimed is:

1. A method of determining whether to treat soybeans for soybean aphids, the method comprising:
collecting at least one image of a soybean canopy using one or more remote sensing instruments including a multispectral camera;
processing the image into spectral reflectance data and selecting from the spectral reflectance data optimal spectral wavelength bands;
classifying the selected reflectance data into one of a plurality of classification groupings using a machine learned classification model, wherein the classification groupings comprise at least a first group that corresponds with a count of aphids that is below an economic threshold count of 250 aphids per plant and a second group that corresponds with a count of aphids that is above the economic threshold count of 250 aphids per plant; and
determining whether to treat or not treat the soybean canopy for aphids based on the classification of the reflectance data into one of the class groupings.

2. The method of claim 1 , wherein selecting from the spectral reflectance data optimal spectral wavelength bands comprises selecting from the spectral reflectance data spectral wavelength bands that are indicative of soybean aphid-induced stress.

3. The method of claim 1 , wherein selecting from the spectral reflectance data optimal spectral wavelength bands comprises selecting from the spectral reflectance data spectral wavelength bands that are indicative of soybean aphid-induced stress and spectral wavelength bands that eliminate false positives for soybean aphid infestation.

4. The method of claim 1 , wherein the optimal spectral wavelength bands comprise red edge and near red infrared (NIR) wavelength bands.

5. The method of claim 1 , wherein the machine learned classification model is trained with spectral reflectance data in the optimal spectral wavelength bands and corresponding actual aphid counts.

6. The method of claim 5 , wherein the optimal spectral wavelength bands that are used to train the machine learned classification model comprise spectral wavelength bands that are indicative of soybean aphid-induced stress including at least 780 nm and 1010 nm or both.

7. The method of claim 5 , wherein the optimal spectral wavelength bands that are used to train the machine learned classification model comprise spectral wavelength bands that are indicative of soybean aphid-induced stress and spectral wavelength bands that eliminate false positives for soybean aphid infestation including at least 711 nm, 789 nm, 919 nm, 1,010 nm and 1,044 nm or combinations thereof.

8. A computer system comprising:
a camera configured to collect at least one image of a soybean canopy;
a memory storing spectral reflectance data processed from the at least one image of the soybean canopy; and
a processor executing instructions to perform steps comprising:
selecting from the spectral reflectance data optimal spectral wavelength bands;
classifying the selected reflectance data into one of a plurality of classification groupings using a machine learned classification model trained with spectral reflectance data and corresponding actual aphid counts, wherein the processor classifies the selected reflectance data into at least a first group that corresponds with a count of aphids that is below an economic threshold count of 250 aphids per plant and a second group that corresponds with a count of aphids that is above the economic threshold count of 250 aphids per plant; and
determining whether to treat or not treat the soybean canopy for aphids based on the classification of the reflectance data into one of the class groupings.

9. The computer system of claim 8 , wherein the optimal spectral wavelength bands comprise spectral wavelength bands that are indicative of soybean aphid-induced stress.

10. The computer system of claim 8 , wherein the optimal spectral wavelength bands comprise spectral wavelength bands that are indicative of soybean aphid-induced stress and spectral wavelength bands that eliminate false positives for soybean aphid infestation.

11. The computer system of claim 8 , wherein the optimal spectral wavelength bands comprise red edge and near red infrared (NIR) wavelength bands.

12. The computer system of claim 8 , wherein the spectral reflectance data that the machine learned classification model is trained with comprises spectral reflectance data that is in the optimal spectral wavelength bands.

13. A method comprising:
collecting at least one image of an area of a soybean field using a camera;
processing the at least one image into spectral reflectance data;
selecting from the spectral reflectance data optimal spectral wavelength bands; classifying the selected reflectance data into one of a plurality of classification groupings using a machine learned classification model trained with spectral reflective data in the select optimal spectral wavelength bands and corresponding actual aphid counts, wherein the selected reflectance data is classified into at least a first group that corresponds with a count of aphids that is below an economic threshold count of 250 aphids per plant and a second group that corresponds with a count of aphids that is above the economic threshold count of 250 aphids per plant; and
determining whether to treat or not treat the soybean canopy for aphids based on the classification of the reflectance data into one of the class groupings.

14. The method of claim 13 , wherein the optimal spectral wavelength bands comprise red edge and near red infrared (NIR) wavelength bands.

15. The method of claim 13 , wherein the optimal spectral wavelength bands that are used to train the machine learned classification model comprise spectral wavelength bands that are indicative of soybean aphid-induced stress including at least 780 nm and 1010 nm or both.

16. The method of claim 13 , wherein the optimal spectral wavelength bands that are used to train the machine learned classification model comprise spectral wavelength bands that are indicative of soybean aphid-induced stress and spectral wavelength bands that eliminate false positives for soybean aphid infestation including at least 711 nm, 789 nm, 919 nm, 1,010 nm and 1,044 nm or combinations thereof.

Assignees

• Univ Minnesota

Inventors

• Koch Robert L
• Macrae Ian V
• Marston Zachary Peter Dragan
• Mulla David

Classifications

• G01N33/025 Primary

  Fruits or vegetables · CPC title

• G06V20/188

  Vegetation · CPC title

• G06F18/217

  Validation; Performance evaluation; Active pattern learning techniques · CPC title

• G06F18/24

  Classification techniques · CPC title

• G06F18/214

  Generating training patterns; Bootstrap methods, e.g. bagging or boosting · CPC title