SYSTEMS AND METHODS FOR TREATING CROP DISEASES IN GROWING SPACES

1 . A computer-implemented method for directing crop disease treatments to plots, the computer-implemented method comprising:

receiving, by a computing device, a request for a crop disease prediction related to treatment of a target plot for one or more crop disease, the request including crop disease type data and location data relating to the target plot, the crop disease type data including multiple identifiers each associated with a different one of multiple crop disease types;

accessing, by the computing device, a multiple disease joint model consistent with the location data;

determining, by the computing device, via the multiple disease joint model, a first disease likelihood output and a second disease likelihood output based on at least the location data, the first disease likelihood output and the second disease likelihood output each associated with a different one of the multiple disease types;

generating, by the computing device, a treatment recommendation based on the first disease likelihood output and the second disease likelihood output of the multiple disease joint model; and

directing, by the computing device, application of at least one treatment to the target plot, based on the treatment recommendation output.

2 . The computer-implemented method of claim 1 , wherein the multiple disease joint model includes at least one of a coregionalization architecture with a separable covariance function, a linear regression model, a neural network regression model, a neural network covariance function, a vector autoregression architecture, a multi-task learning architecture, a shared smoothing penalties architecture, and a Gaussian process model.

3 . The computer-implemented method of claim 2 , wherein the multiple disease joint model includes a coregionalization architecture to jointly model occurrence probabilities and/or disease severity probabilities of multiple crop disease types, where model data is shared across the multiple crop disease types, locations of multiple plots, and multiple observation dates, using a separable covariance function.

4 . The computer-implemented method of claim 3 , wherein the multiple disease joint model includes at least one of a linear regression mean function and a neural network regression mean function for relating input parameters characterizing the target plot to likelihood probabilities of multiple crop disease types.

5 . The computer-implemented method of claim 1 , wherein:

the multiple disease joint model includes a Gaussian process model that interpolates data across plot locations and crop disease observation dates; and

an output of the multiple disease joint model includes an output of a neural network combined with a smooth Gaussian process.

6 . The computer-implemented method of claim 1 , further comprising training the multiple disease joint model, based on historical data associated with multiple plots and multiple crop disease types.

7 . The computer-implemented method of claim 6 , wherein inputs for training the multiple disease joint model include, for each of the multiple plots:

a location of the plot;

a presence or severity of multiple crop diseases at the plot; and

a date of observation of crops of the plot to determine the presence or severity of the multiple crop diseases.

8 . The computer-implemented method of claim 7 , wherein the inputs for training the multiple disease joint model include, for each of the multiple plots, at least one of soil information of the plot, field topology information of the plot, weather information associated with the plot, field management practice information associated with the plot, and hybrid/genetic seed information associated with crops on the plot.

9 . The computer-implemented method of claim 6 , wherein training the model includes:

accessing data specific to a region of the target plot;

manipulating, by the computing device, the accessed data; and

training, by the computing device, the multiple disease joint model based on at least a portion of the manipulated data.

10 . The computer-implemented method of claim 1 , further comprising treating the target plot with the treatment in response to the treatment recommendation.

11 . The computer-implemented method of claim 10 , wherein treating the target plot with the treatment includes applying the treatment to crops in the target plot.

12 . The computer-implemented method of claim 10 , further comprising:

receiving, at a communication device of a user associated with the target plot, the treatment recommendation; and

causing operation of one or more agricultural apparatuses at the target plot to apply the treatment to the crops in the target plot.

13 . The computer-implemented method of claim 1 , further comprising providing a forecasted disease risk map and/or a time series view, via an application and/or website, the map and/or view indicative of the first disease likelihood output and the second disease likelihood output.

14 . The computer-implemented method of claim 1 , wherein the request for the crop disease prediction is specific to a first crop type, the method further comprising:

receiving, by the computing device, a second request for a second crop disease prediction related to treatment of a second crop type at the target plot;

determining, by the computing device, via the multiple disease joint model, a third disease likelihood output and a fourth disease likelihood output, based on at least the location data and the second crop type, the third disease likelihood output and the fourth disease likelihood output each associated with the second crop type and a different one of the multiple disease types; and

generating, by the computing device, a second treatment recommendation based on the third disease likelihood output and the fourth disease likelihood output.

15 . The computer-implemented method of claim 1 , further comprising:

receiving, by the computing device, a second request for a second crop disease prediction related to treatment of a second target plot;

determining, by the computing device, via the multiple disease joint model, a third disease likelihood output and a fourth disease likelihood output, based on at least location data of the second target plot, the third disease likelihood output and the fourth disease likelihood output each associated with the second target plot and a different one of the multiple disease types; and

generating, by the computing device, a second treatment recommendation based on the third disease likelihood output and the fourth disease likelihood output.

16 . A system for directing crop disease treatments to plots, the system comprising at least one computing device configured to:

receive a request for a crop disease prediction related to treatment of a target plot for one or more crop disease, the request including crop disease type data and location data relating to the target plot, the crop disease type data including multiple identifiers each associated with a different one of multiple crop disease types;

access a multiple disease joint model consistent with the location data;

determine, via the multiple disease joint model, a first disease likelihood output and a second disease likelihood output based on at least the location data, the first disease likelihood output and the second disease likelihood output each associated with a different one of the multiple disease types;

generate a treatment recommendation based on the first disease likelihood output and the second disease likelihood output of the multiple disease joint model

17 . The system of claim 16 , further comprising at least one agricultural machine configured to apply the treatment to crops in the target plot.

18 . A non-transitory computer-readable storage medium including computer-executable instructions for use in directing crop disease treatments to plots, which when executed by at least one processor, cause the at least one processor to:

receive a request for a crop disease prediction related to treatment of a target plot for one or more crop disease, the request including crop disease type data and location data relating to the target plot, the crop disease type data including multiple identifiers each associated with a different one of multiple crop disease types;

access a multiple disease joint model consistent with the location data;

determine, via the multiple disease joint model, a first disease likelihood output and a second disease likelihood output based on at least the location data, the first disease likelihood output and the second disease likelihood output each associated with a different one of the multiple disease types;

generate a treatment recommendation based on the first disease likelihood output and the second disease likelihood output of the multiple disease joint model.

19 . The non-transitory computer-readable storage medium of claim 18 , wherein the multiple disease joint model includes a coregionalization architecture to jointly model occurrence probabilities and/or disease severity probabilities of multiple crop disease types, where model data is shared across the multiple crop disease types, locations of multiple plots, and multiple observation dates, using a separable covariance function.

20 . The non-transitory computer-readable storage medium of claim 18 , wherein:

the multiple disease joint model includes a Gaussian process model that interpolates data across plot locations and crop disease observation dates; and

an output of the multiple disease joint model includes an output of a neural network combined with a smooth Gaussian process.