Back to results
US 12,373,849 B2Granted
Systems and methods for validating planting of trees
AssigneeQuanata, LLC
InventorKenneth Jason Sanchez
Life of this patent
Estimating the term…
Abstract
Method and system for validating planting of trees. For example, the method includes receiving a first image depicting a tree, receiving a second image depicting an environment where the tree is to be planted, receiving a third image depicting the tree having been planted in the environment, selecting and encoding a first patch of the first image that depicts the tree, selecting and encoding a second patch of the second image that depicts the environment, selecting and encoding a third patch of the third image that depicts both the tree and the environment, and comparing the encoded patches to determine whether the tree has been planted in the environment.
Claims20
A computer-implemented method comprising:
- selecting, by a computing device, a first patch of a first image that depicts a part of a tree or a second patch of a second image that depicts a part of an environment, and a third patch of a third image that depicts the part of the tree and the part of the environment;
- encoding, by the computing device, the first patch of the first image to generate first encoded tree data or the second patch of the second image to generate first encoded environment data, and the third patch of the third image to generate third encoded tree data or third encoded environment data, wherein the encoding the first patch or the second patch involves learning to generate the first encoded tree data or the first encoded environment data, respectively, based upon a model, wherein the model is an artificial neural network;
- analyzing, by the computing device, the first patch or the second patch, wherein the analyzing the first patch or the second patch includes performing one or more of feature extraction or applying pattern recognition;
- using, by the computing device, an autoencoder to generate the first encoded tree data or the first encoded environment data, wherein the autoencoder includes a reduction layer for inputs, one or more intermediate layers, and a reconstruction layer for outputs;
- comparing, by the computing device, the first encoded tree data with the third encoded tree data or the first encoded environment data with the third encoded environment data to generate one or more comparison data;
- and
- determining, by the computing device, whether the tree is in the environment based at least in part upon the one or more comparison data.
4 more elements
A computing device comprising:
- one or more processors;
- and
- a memory storing instructions that, when executed by the one or more processors, cause the one or more processors to:
- select a first patch of a first image that depicts a part of a tree or a second patch of a second image that depicts a part of an environment, and a third patch of a third image that depicts the part of the tree and the part of the environment;
- encode the first patch of the first image to generate first encoded tree data or the second patch of the second image to generate first encoded environment data, and the third patch of the third image to generate third encoded tree data or third encoded environment data, wherein encoding the first patch or the second patch involves learning to generate the first encoded tree data or the first encoded environment data, respectively, based upon a model, wherein the model is an artificial neural network;
- analyze the first patch or the second patch, wherein analyzing the first patch or the second patch includes performing one or more of feature extraction or applying pattern recognition;
- use an autoencoder to generate the first encoded tree data or the first encoded environment data, wherein the autoencoder includes a reduction layer for inputs, one or more intermediate layers, and a reconstruction layer for outputs;
- compare the first encoded tree data with the third encoded tree data or the first encoded environment data with the third encoded environment data to generate one or more comparison data;
- and
- determine whether the tree is in the environment based at least in part upon the one or more comparison data.
7 more elements
A non-transitory computer-readable medium storing instructions, the instructions when executed by one or more processors of a computing device, cause the computing device to:
- select a first patch of a first image that depicts a part of a tree or a second patch of a second image that depicts a part of an environment, and a third patch of a third image that depicts the part of the tree and the part of the environment;
- encode the first patch of the first image to generate first encoded tree data or the second patch of the second image to generate first encoded environment data, and the third patch of the third image to generate third encoded tree data or third encoded environment data, wherein encoding the first patch or the second patch involves learning to generate the first encoded tree data or the first encoded environment data, respectively, based upon a model, wherein the model is an artificial neural network;
- analyze the first patch or the second patch, wherein analyzing the first patch or the second patch includes performing one or more of feature extraction or applying pattern recognition;
- use an autoencoder to generate the first encoded tree data or the first encoded environment data, wherein the autoencoder includes a reduction layer for inputs, one or more intermediate layers, and a reconstruction layer for outputs;
- compare the first encoded tree data with the third encoded tree data or the first encoded environment data with the third encoded environment data to generate one or more comparison data;
- and
- determine whether the tree is in the environment based at least in part upon the one or more comparison data.
4 more elements
Dependencies are inferred from the claim text.