# US 12,213,395 B2 · Monitoring and intelligence generation for farm field

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 12,213,395 B2
- Document ID: US12213395B2
- Status: Granted
- Kind: B2
- Application number: 17/447,984
- Filed: 2021-09-17
- Published: 2023-03-23
- Granted: 2025-02-04
- Earliest claimed: 2021-09-17
- Assignees: 6th Grain Corporation
- Applicants: 6th Grain Corporation
- Inventors: Molly Elizabeth Brown; Min Feng; Vladimir Eskin
- Web page: https://patentgrep.com/patent/US12213395B2

## CPC

- A01B 79/005: {Precision agriculture}
- G06N 3/0455: Auto-encoder networks; Encoder-decoder networks
- G06N 3/0464: Convolutional networks [CNN, ConvNet]
- G06N 3/09: Supervised learning
- G06N 5/01: Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound
- G06N 20/00: Machine learning
- G06Q 50/02: Agriculture; Fishing; Forestry; Mining
- G06V 10/758: {Involving statistics of pixels or of feature values, e.g. histogram matching}
- G06V 20/13: Satellite images
- G06V 20/188: {Vegetation}
- G06N 3/082: modifying the architecture, e.g. adding, deleting or silencing nodes or connections
- A01B: SOIL WORKING IN AGRICULTURE OR FORESTRY; PARTS, DETAILS, OR ACCESSORIES OF AGRICULTURAL MACHINES OR IMPLEMENTS, IN GENERAL (making or covering furrows or holes for sowing, planting or manuring A01C5/00; mowers convertible to soil working apparatus or capable of soil working A01D42/04; mowers combined with soil working implements A01D43/12; soil working for engineering purposes E01, E02, E21)
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06Q: INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING

## Abstract

Various embodiments described herein provide monitoring and intelligence generation for one or more farm fields by: using remote sensing to monitor progress of a developing crop; comparing a user's field to another field in the area; providing cropped area extent estimates for a current season; using one or more disease risk models to determine disease risk (or disease pressure) with respect to a field; or some combination thereof.

## Claims

### Claim 1

A method comprising:
accessing, by a hardware processor, tile data describing a plurality of tile images for a geographic region over a period of time, the tile data comprising pixels and being generated based on satellite remote sensing data;
generating, by the hardware processor and based on the tile data, farm field pixel data for a farm field of interest, the farm field pixel data describing a plurality of pixels for the farm field of interest over the period of time;
generating, by the hardware processor and based on the farm field pixel data, vegetation index data for the farm field of interest by, for each individual pixel in the plurality of pixels:
generating an initial time series of vegetation index values for the individual pixel over the period of time; and
generating a reconstructed time series of vegetation index values for the individual pixel over the period of time by using a first machine learning model to generate a predicted vegetation index value for each vegetation index value in the initial time series of vegetation index values, the first machine learning model comprising a neural network model trained to reconstruct vegetation index values of all pixels of a field;
generating, by the hardware processor, a crop growth curve for the farm field of interest based on the vegetation index data;
generating, by the hardware processor, a crop growth band based on historical data that describes vegetation index values for one or more other farm fields in the geographic region;
causing, by the hardware processor, presentation of the crop growth band and the crop growth curve on a client graphical user interface on a client device, the crop growth band and the crop growth curve being presented relative to each other for comparison;
generating, using a second machine learning model, a simulated yield value for the field of interest based on the crop growth band; and
generating, by the hardware processor, one or more alerts based on the simulated yield value for the field of interest, the one or more alerts being reviewable by a user, at least one alert of the one or more alerts providing actionable information to the user for adjusting management of the field of interest.

### Claim 2 (depends on 1)

The method of claim 1, wherein the generating of the crop growth curve for the farm field of interest based on the vegetation index data comprises:
generating a time series of average vegetation index values by averaging together all vegetation index values, for each time point in a range of time within the period of time, from all of the reconstructed time series of vegetation index values; and
generating the crop growth curve based on the time series of average vegetation index values.

### Claim 3 (depends on 1)

The method of claim 1, wherein the generating of the crop growth band based on the historical data comprises:
selecting the one or more other farm fields from a plurality of farm fields in the geographic region;
selecting, from the one or more other farm fields, a subset of farm fields that each have a yield that is higher than a lower percentile threshold and lower than a higher percentile threshold;
generating a median vegetation index curve for those fields, in the subset of farm fields, that have a yield that is higher than a upper percentile threshold; and
generating the crop growth band based on the median vegetation index curve.

### Claim 4 (depends on 2)

The method of claim 2, wherein the method comprises:
determining a regional yield estimate from the crop growth band;
determining a daily accumulated vegetation index value based on the time series of average vegetation index values; and
determining a daily change of vegetation index value based on the time series of average vegetation index values, the generating of the simulated yield value for the field of interest based on the crop growth band comprising generating the simulated yield value based on the regional yield estimate, the daily accumulated vegetation index, and the daily change of vegetation index value.

### Claim 5 (depends on 1)

The method of claim 1, wherein the machine learning model as trained is configured to receive an input vegetation index value and generate an output vegetation index value that represents a correction of the input vegetation index value or a replacement of the input vegetation index value.

### Claim 6 (depends on 1)

The method of claim 1, wherein a Mean Square Error (MSE) process is used to generate a learning metric for the machine learning model.

### Claim 7 (depends on 1)

The method of claim 1, wherein each individual vegetation index value, in an individual time series for the individual pixel, being calculated based on a Normalized Difference Vegetation Index value computed for the individual pixel at a different time point within the period of time.

### Claim 8 (depends on 7)

The method of claim 7, wherein the individual vegetation index value is calculated based on the Normalized Difference Vegetation Index value computed for the individual pixel by:
truncating the Normalized Difference Vegetation Index value for the individual pixel from 0 to 1; and
multiplying the truncated Normalized Difference Vegetation Index by a predetermined factor of 100.

### Claim 9

A non-transitory computer-readable storage medium including instructions that, when processed by a hardware processor, causes the hardware processor to perform operations comprising:
accessing tile data describing a plurality of tile images for a geographic region over a period of time, the tile data comprising pixels and being generated based on satellite remote sensing data;
generating, based on the tile data, farm field pixel data for a farm field of interest, the farm field pixel data describing a plurality of pixels for the farm field of interest over the period of time;
generating, based on the farm field pixel data, vegetation index data for the farm field of interest by, for each individual pixel in the plurality of pixels:
generating an initial time series of vegetation index values for the individual pixel over the period of time; and
generating a reconstructed time series of vegetation index values for the individual pixel over the period of time, by using a first machine learning model to generate a predicted vegetation index value for each vegetation index value in the initial time series of vegetation index values, the first machine learning model comprising a neural network model trained to reconstruct vegetation index values of all pixels of a field;
generating a crop growth curve for the farm field of interest based on the vegetation index data;
generating a crop growth band based on historical data that describes vegetation index values for one or more other farm fields in the geographic region;
causing presentation of the crop growth band and the crop growth curve on a client graphical user interface on a client device, the crop growth band and the crop growth curve being presented relative to each other for comparison;
generating, using a second machine learning model, a simulated yield value for the field of interest based on the crop growth band; and
generating one or more alerts based on the simulated yield value for the field of interest, the one or more alerts being reviewable by a user, at least one alert of the one or more alerts providing actionable information to the user for adjusting management of the field of interest.

### Claim 10 (depends on 9)

The non-transitory computer-readable storage medium of claim 9, wherein the generating of the crop growth curve for the farm field of interest based on the vegetation index data comprises:
generating a time series of average vegetation index values by averaging together all vegetation index values, for each time point in a range of time within the period of time, from all of the reconstructed time series of vegetation index values; and
generating the crop growth curve based on the time series of average vegetation index values.

### Claim 11 (depends on 9)

The non-transitory computer-readable storage medium of claim 9, wherein the generating of the crop growth band based on the historical data comprises:
selecting the one or more other farm fields from a plurality of farm fields in the geographic region;
selecting, from the one or more other farm fields, a subset of farm fields that each have a yield that is higher than a lower percentile threshold and lower than a higher percentile threshold;
generating a median vegetation index curve for those fields, in the subset of farm fields, that have a yield that is higher than an upper percentile threshold; and
generating the crop growth band based on the median vegetation index curve.

### Claim 12 (depends on 10)

The non-transitory computer-readable storage medium of claim 10, wherein the operations comprise:
determining a regional yield estimate from the crop growth band;
determining a daily accumulated vegetation index value based on the time series of average vegetation index values; and
determining a daily change of vegetation index value based on the time series of average vegetation index values, the generating of the simulated yield value for the field of interest based on the crop growth band comprising generating the simulated yield value based on the regional yield estimate, the daily accumulated vegetation index, and the daily change of vegetation index value.

### Claim 13 (depends on 9)

The non-transitory computer-readable storage medium of claim 9, wherein the machine learning model as trained is configured to receive an input vegetation index value and generate an output vegetation index value that represents a correction of the input vegetation index value or a replacement of the input vegetation index value.

### Claim 14 (depends on 9)

The non-transitory computer-readable storage medium of claim 9, wherein a Mean Square Error (MSE) process is used to generate a learning metric for the machine learning model.

### Claim 15 (depends on 9)

The non-transitory computer-readable storage medium of claim 9, wherein each individual vegetation index value, in an individual time series for the individual pixel, being calculated based on a Normalized Difference Vegetation Index value computed for the individual pixel at a different time point within the period of time.

### Claim 16 (depends on 15)

The non-transitory computer-readable storage medium of claim 15, wherein the individual vegetation index value is calculated based on the Normalized Difference Vegetation Index value computed for the individual pixel by:
truncating the Normalized Difference Vegetation Index value for the individual pixel from 0 to 1; and
multiplying the truncated Normalized Difference Vegetation Index by a predetermined factor of 100.

### Claim 17

A system comprising:
a hardware processor; and
a memory storing instructions that, when executed by the processor, cause the hardware processor to perform operations comprising:
accessing tile data describing a plurality of tile images for a geographic region over a period of time, the tile data comprising pixels and being generated based on satellite remote sensing data;
generating, based on the tile data, farm field pixel data for a farm field of interest, the farm field pixel data describing a plurality of pixels for the farm field of interest over the period of time;
generating, based on the farm field pixel data, vegetation index data for the farm field of interest by, for each individual pixel in the plurality of pixels:
generating an initial time series of vegetation index values for the individual pixel over the period of time; and
generating a reconstructed time series of vegetation index values for the individual pixel over the period of time, by using a first machine learning model to generate a predicted vegetation index value for each vegetation index value in the initial time series of vegetation index values, the first machine learning model comprising a neural network model trained to reconstruct vegetation index values of all pixels of a field;
generating a crop growth curve for the farm field of interest based on the vegetation index data;
generating a crop growth band based on historical data that describes vegetation index values for one or more other farm fields in the geographic region;
causing presentation of the crop growth band and the crop growth curve on a client graphical user interface on a client device, the crop growth band and the crop growth curve being presented relative to each other for comparison;
generating, using a second machine learning model, a simulated yield value for the field of interest based on the crop growth band; and
generating one or more alerts based on the simulated yield value for the field of interest, the one or more alerts being reviewable by a user, at least one alert of the one or more alerts providing actionable information to the user for adjusting management of the field of interest.

### Claim 18 (depends on 17)

The system of claim 17, wherein the generating of the crop growth curve for the farm field of interest based on the vegetation index data comprises:
generating a time series of average vegetation index values by averaging together all vegetation index values, for each time point in a range of time within the period of time, from all of the reconstructed time series of vegetation index values; and
generating the crop growth curve based on the time series of average vegetation index values.

### Claim 19 (depends on 17)

The system of claim 17, wherein the generating of the crop growth band based on the historical data comprises:
selecting the one or more other farm fields from a plurality of farm fields in the geographic region;
selecting, from the one or more other farm fields, a subset of farm fields that each have a yield that is higher than a lower percentile threshold and lower than a higher percentile threshold;
generating a median vegetation index curve for those fields, in the subset of farm fields, that have a yield that is higher than an upper percentile threshold; and
generating the crop growth band based on the median vegetation index curve.