Spot weed detection and treatment within a field of view in accordance with machine learning training
A weed spot-spraying system is described for carrying out a spot-based weed spraying method based upon a classification value rendered from a sub-field image in accordance with a machine learning-based trained model applied by a processor to the sub-field image. The system includes a camera; a spray nozzle assembly including a spray nozzle; and a processor. The method carried out by the system includes acquiring, by the camera, a full field of view image of a crop floor. The method further includes extracting, from the full field of view image, a sub-field image corresponding to the spray nozzle positioned to provide a spray field extending over a part of the crop floor depicted in the sub-field image; rendering, by the processor in accordance with the machine learning-based trained model, a classification for the sub-field image; and selectively activating the spray nozzle in accordance with the classification for the sub-field image.
1 . A method for carrying out spot-based weed spraying based upon a classification value rendered from a sub-field image in accordance with a machine learning-based trained model applied by a processor to the sub-field image, the method comprising:
acquiring, by a camera, a full field of view image of a crop floor;
extracting, from the full field of view image, a sub-field image corresponding to a spray nozzle positioned to provide a spray field extending over a part of the crop floor depicted in the sub-field image;
rendering, by the processor in accordance with the machine learning-based trained model, a classification for the sub-field image; and
selectively activating the spray nozzle in accordance with the classification for the sub-field image,
wherein the rendering is carried out by a neural network incorporated into the processor, and
wherein the neural network has an array of inputs arranged in accordance with pixel dimensions of the sub-field image corresponding to a spray nozzle positioned to provide a spray field.
2 . The method of claim 1 , wherein the neural network has a quantity of outputs corresponding to a set of potential classifications for the sub-field image.
3 . The method of claim 1 , wherein the method is carried out on a system including a plurality of spray nozzles having overlapping spray fields between two adjacent ones of the plurality of spray nozzles.
4 . The method of claim 1 , wherein the camera is mounted upon a same physical mounting structure as the spray nozzle.
5 . The method of claim 1 , wherein the sub-field image is rectangular shaped.
6 . A method for carrying out spot-based weed spraying based upon a classification value rendered from a sub-field image in accordance with a machine learning-based trained model applied by a processor to the sub-field image, the method comprising:
acquiring, by a camera, a full field of view image of a crop floor;
extracting, from the full field of view image, a sub-field image corresponding to a spray nozzle positioned to provide a spray field extending over a part of the crop floor depicted in the sub-field image;
rendering, by the processor in accordance with the machine learning-based trained model, a classification for the sub-field image; and
selectively activating the spray nozzle in accordance with the classification for the sub-field image,
wherein the sub-field image is a rectangle rendered from an initial trapezoid image sub-field captured by a forward-looking camera.
7 . The method of claim 1 , wherein the selectively activating the spray nozzle comprises leaving an activated spray nozzle in an on state for a period of time that relates to a speed with which the spay nozzle travels along a line.
8 . The method of claim 1 , further comprising adapting an on-time of the spray nozzle in accordance with a machine speed.
9 . The method of claim 1 , further comprising providing a sample sub-field image and a corresponding verified classification for the sample-subfield image to a training facility for the machine learning-based trained model.
10 . The method of claim 9 , wherein the sample sub-field image is provided in real-time during the spot spraying operation.
11 . The method of claim 10 , wherein the providing the sample sub-field image to a reviewer for verification of an assigned classification is performed on an automated basis.
12 . The method of claim 11 , wherein the automated basis comprises applying a confidence value for the classification of the sample sub-field image to a threshold confidence value.
13 . The method of claim 1 , wherein the rendering is performed in a cloud-based server system.
14 . The method of claim 1 , wherein the method further comprises providing the rendered classification and associated sub-field image to a reviewer for annotation.
15 . The method of claim 1 , wherein the rendering is carried out in accordance with a foundation model.
16 . The method of claim 15 , wherein the foundation model is provided with an associated training coverage map indicating an extent of training images of particular types for which machine leaning training has been performed.
17 . A weed spot-spraying system configured to carry out a spot-based weed spraying method based upon a classification value rendered from a sub-field image in accordance with a machine learning-based trained model applied by a processor to the sub-field image, comprising:
a camera;
a spray nozzle assembly including a spray nozzle; and
a processor;
wherein the spot-based weed spraying method comprises:
acquiring, by the camera, a full field of view image of a crop floor;
extracting, from the full field of view image, a sub-field image corresponding to the spray nozzle positioned to provide a spray field extending over a part of the crop floor depicted in the sub-field image;
rendering, by the processor in accordance with the machine learning-based trained model, a classification for the sub-field image; and
selectively activating the spray nozzle in accordance with the classification for the sub-field image,
wherein the rendering is carried out by a neural network incorporated into the processor, and
wherein the neural network has an array of inputs arranged in accordance with pixel dimensions of the sub-field image corresponding to a spray nozzle positioned to provide a spray field.
18 . The weed spot-spraying system of claim 17 , wherein the system further comprises a network communication interface configured to communicate with a networked facility to provide training messages comprising:
a sub-field image instance;
metadata describing an environment within which the sub-field image instance was acquired by the camera, and
a confirmed characterization of the sub-field image.
19 . The weed spot-spraying system of claim 18 , wherein the neural network has a quantity of outputs corresponding to a set of potential classifications for the sub-field image.
20 . A weed spot-spraying system configured to carry out a spot-based weed spraying method based upon a classification value rendered from a sub-field image in accordance with a machine learning-based trained model applied by a processor to the sub-field image, comprising:
a camera;
a spray nozzle assembly including a spray nozzle; and
a processor;
wherein the spot-based weed spraying method comprises:
acquiring, by the camera, a full field of view image of a crop floor;
extracting, from the full field of view image, a sub-field image corresponding to the spray nozzle positioned to provide a spray field extending over a part of the crop floor depicted in the sub-field image;
rendering, by the processor in accordance with the machine learning-based trained model, a classification for the sub-field image; and
selectively activating the spray nozzle in accordance with the classification for the sub-field image,
wherein the sub-field image is a rectangle rendered from an initial trapezoid image sub-field captured by a forward-looking camera.