Detecting and measuring the size of clods and other soil features from imagery

Patent metadata

Field	Value
Publication number	US-2020005474-A1
Application number	US-201816020097-A
Country	US
Kind code	A1
Filing date	Jun 27, 2018
Priority date	Jun 27, 2018
Publication date	Jan 2, 2020
Grant date	—

Abstract

The present disclosure provides systems and methods that measure soil roughness in a field from imagery of the field. In particular, the present subject matter is directed to systems and methods that include or otherwise leverage a machine-learned clod detection model to determine a soil roughness value for a portion of a field based at least in part on imagery of such portion of the field captured by an imaging device. For example, the imaging device can be a camera positioned in a downward-facing direction and physically coupled to a work vehicle or an implement towed by the work vehicle through the field.

First claim

What is claimed is:

1 . A computing system, comprising:
one or more processors; and one or more non-transitory computer-readable media that collectively store:
a machine-learned clod detection model configured to receive imagery and to process the imagery to detect soil clods depicted by the imagery; and
instructions that, when executed by the one or more processors, configure the computing system to perform operations, the operations comprising:
obtaining image data that depicts a portion of a field;
inputting the image data into the machine-learned clod detection model;
receiving one or more bounding shapes as an output of the machine-learned clod detection model, wherein each of the one or more bounding shapes indicates a respective location of a respective soil clod detected by the machine-learned clod detection model in the image data; and
determining a soil roughness value for at least the portion of the field based at least in part on the one or more bounding shapes.

2 . The computing system of claim 1 , wherein:
the computing system is physically located on at least one of a work vehicle or an implement towed by the work vehicle; obtaining the image data of the field comprises obtaining image data captured by an imaging device that is physically located on at least one of the work vehicle or the implement towed by the work vehicle; and the operations further comprise controlling an operation of at least one of the work vehicle or the implement towed by the work vehicle based at least in part on the soil roughness value determined from the image data.

3 . The computing system of claim 1 , wherein the machine-learned clod detection model comprises a Haar Cascade Classifier.

4 . The computing system of claim 1 , wherein determining the soil roughness value for at least the portion of the field based at least in part on the one or more bounding shapes comprises:
determining a respective size of each of the one or more bounding shapes; and determining the soil roughness value for at least the portion of the field based at least in part on the respective sizes of the one or more bounding shapes.

5 . The computing system of claim 4 , wherein the respective size of each bounding shape comprises one or more of a height, a width, or an area of such bounding shape.

6 . The computing system of claim 5 , wherein determining the soil roughness value for at least the portion of the field based at least in part on the respective sizes of the one or more bounding shapes comprises:
determining one or more statistics descriptive of the respective sizes of the one or more bounding shapes; and determining the soil roughness value for at least the portion of the field based at least in part on the one or more statistics.

7 . The computing system of claim 6 , wherein the one or more statistics comprise one or more of a mean size, a median size, a standard deviation of the respective sizes, or a distribution histogram of sizes.

8 . The computing system of claim 6 , wherein determining the soil roughness value for at least the portion of the field based at least in part on one or more statistics comprises:
updating one or more running averages based at least in part on the one or more statistics; and determining the soil roughness value for at least the portion of the field based at least in part on the one or more running averages.

9 . The computing system of claim 8 , wherein the one or more running averages comprise one or more of:
a recursive least squares filter; a least mean squares filter; an autoregressive-moving-average model; or a Kalman filter.

10 . The computing system of claim 1 , wherein the image data comprises a plurality of image frames.

11 . The computing system of claim 1 , wherein the image data comprises a single image frame and wherein the operations further comprise:
obtaining a plurality of image frames that depict respective portions of the field; performing the operations recited in claim 1 for each of the plurality of image frames to determine a plurality of frame-wise soil roughness values respectively for the plurality of image frames; and determining a total soil roughness value for the field based at least in part on the plurality of frame-wise soil roughness values.

12 . A computer-implemented method, comprising:
obtaining, by a computing system comprising one or more computing devices, image data that depicts a portion of a field; inputting, by the computing system, the image data into machine-learned clod detection model configured to receive imagery and to process the imagery to detect soil clods depicted by the imagery; receiving, by the computing system, one or more bounding shapes as an output of the machine-learned clod detection model, wherein each of the one or more bounding shapes indicates a respective location of a respective soil clod detected by the machine-learned clod detection model in the image data; and determining, by the computing system, a soil roughness value for at least the portion of the field based at least in part on the one or more bounding shapes.

13 . The computer-implemented method of claim 12 , wherein:
the computing system is physically located on at least one of a work vehicle or an implement towed by the work vehicle; obtaining, by the computing system, the image data of the field comprises obtaining, by the computing system, image data captured by an imaging device that is physically located on at least one of the work vehicle or the implement towed by the work vehicle; and the method further comprises controlling an operation of at least one of the work vehicle or the implement towed by the work vehicle based at least in part on the soil roughness value determined from the image data.

14 . The computer-implemented method of claim 12 , wherein the machine-learned clod detection model comprises a Haar Cascade Classifier.

15 . The computer-implemented method of claim 12 , wherein determining, by the computing system, the soil roughness value for at least the portion of the field based at least in part on the one or more bounding shapes comprises:
determining, by the computing system, a respective size of each of the one or more bounding shapes; and determining, by the computing system, the soil roughness value for at least the portion of the field based at least in part on the respective sizes of the one or more bounding shapes.

16 . The computer-implemented method of claim 15 , wherein the respective size of each bounding shape comprises one or more of a height, a width, or an area of such bounding shape.

17 . The computer-implemented method of claim 15 , wherein determining, by the computing system, the soil roughness value for at least the portion of the field based at least in part on the respective sizes of the one or more bounding shapes comprises:
determining, by the computing system, one or more statistics descriptive of the respective sizes of the one or more bounding shapes; and determining, by the computing system, the soil roughness value for at least the portion of the field based at least in part on the one or more statistics.

18 . The computer-implemented method of claim 17 , wherein determining, by the computing system, the soil roughness value for at least the portion of the field based at least in part on one or more statistics comprises:
updating, by the computing system, one or more running averages based at least in part on the one or more statistics; and determining, by the computing system, t

Assignees

• Cnh Ind Canada Ltd
• Autonomous Solutions Inc

Inventors

• Ferrari Luca
• Posselius John H
• Henry James W
• Bybee Taylor C

Classifications

• G06T2207/30188

  Vegetation; Agriculture · CPC title

• G06T2207/20081

  Training; Learning · CPC title

• G06T2207/20084

  Artificial neural networks [ANN] · CPC title

• G06V10/46

  Descriptors for shape, contour or point-related descriptors, e.g. scale invariant feature transform [SIFT] or bags of words [BoW]; Salient regional features (colour feature extraction G06V10/56) · CPC title

• G06T7/0012 Primary

  Biomedical image inspection · CPC title