System, apparatus and method of determining refined petiole nutrient values based on leaf spectral data

A non-destructive, real-time method using a portable spectrophotometer and machine learning models for plant petiole nutritional assessment addresses the limitations of traditional destructive testing, enabling timely fertilizer adjustments for improved crop yield and quality.

JP2024169338APending Publication Date: 2024-12-05DALHOUSIE UNIVERSITY
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
JP2024076972
Authority / Receiving Office
JP · JP
Patent Type
Applications
Current Assignee / Owner
Priority Date
2024-04-05
Filing Date
2024-05-10
Publication Date
2024-12-05

AI Technical Summary

Technical Problem

Current methods for determining the nutritional status of plants are destructive, time-consuming, and do not provide real-time results, leading to delayed corrective actions that can be ineffective due to changes in plant nutrient content during the testing process.

Method used

A non-destructive method using a portable spectrophotometer to take spectral measurements of plant leaves, generate leaf spectral data, and apply machine learning models to calculate and provide near real-time nutritional values of petioles, allowing for immediate corrective actions.

Benefits of technology

Enables near real-time determination of petiole nutritional values, facilitating timely adjustments in fertilizer composition and application, thereby improving crop yield and quality.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 2024169338000001_ABST
    Figure 2024169338000001_ABST
Patent Text Reader

Abstract

To provide a system, apparatus and method of determining refined petiole nutrient values based on leaf spectral data.SOLUTION: The method includes: storing leaf spectral data in a memory; using a processor to transform the leaf spectral data into preliminary petiole nutrient values by inputting the leaf spectral data into one or more linear machine learning models trained to provide petiole nutrient value outputs based on leaf spectral data inputs; and using the processor to refine the preliminary petiole nutrient values by inputting the preliminary petiole nutrient values into one or more non-linear machine learning models. The one or more non-linear machine learning models are trained to provide refined petiole nutrient values based on the preliminary petiole nutrient values inputs.SELECTED DRAWING: Figure 1
Need to check novelty before this filing date? Find Prior Art

Description

[Technical field]

[0001] This document relates to systems, devices, and methods for determining the nutritional value of petioles based on leaf spectral data from plant leaves. [Background technology]

[0002]

[0002] Plants require many nutrients to grow and survive. For example, plants may require large amounts of macroelements such as nitrogen (N), phosphorus (P), potassium (K), calcium (Ca), magnesium (Mg), and sulfur (S). Plants may also require small amounts of trace elements such as manganese (Mn), zinc (Zn), iron (Fe), sodium (Na), copper (Cu), aluminum (Al), and boron (B).

[0003]

[0003] In agricultural applications, if a crop plant does not receive the appropriate amount of each necessary nutrient, the crop yield and / or quality of the harvested crop may be affected. For example, if a plant does not receive enough nutrients, the crop yield and / or quality may be reduced. In some cases, an excessive supply of nutrients may also reduce the crop yield and / or quality.

[0004]

[0004] Determining the nutritional status of a plant can allow corrective actions to be implemented to improve crop yield and / or quality. For example, based on the determined nutritional status of the plant, fertilizer composition and amount can be adjusted for one or more specific nutrients. Summary of the Invention

[0005]

[0005] The following summary is provided to introduce the reader to the more detailed discussion that follows. This summary is not intended to limit or define the claimed or yet to be claimed inventions. One or more inventions may reside in any combination or subcombination of the elements or process steps disclosed in any part of this document, including the claims and drawings.

[0006] According to some aspects, a method is provided for non-destructively determining a nutritional value of a petiole from leaves of plants in a crop field in near real time using a portable spectrophotometer. The method may include: (a) performing a non-destructive spectral measurement of leaves of plants in a crop field using the portable spectrophotometer to generate spectral data of the leaves, (b) storing the spectral data of the leaves in a memory, (c) calculating, by a processor, a nutritional value of the petiole based on the stored spectral data of the leaves, and (d) providing a near real time result indicative of the nutritional value of the petiole.

[0007] According to some aspects, an apparatus is provided for non-destructively determining a nutritional value of a petiole from a leaf of a plant in a crop field in near real time. The apparatus may include a memory storing one or more machine learning models trained to provide an output of a nutritional value of the petiole based on an input of spectral data of the leaf, and at least one processor communicatively coupled to the memory. The at least one processor may be collectively configured to receive leaf spectral data generated by a portable spectrophotometer based on a non-destructive spectral measurement of a leaf of a plant in a crop field, store the leaf spectral data in a memory, input the leaf spectral data to one or more machine learning models, receive an output comprising a nutritional value of the petiole determined by the one or more machine learning models based on the spectral data of the leaf, and provide a near real-time result indicative of a nutritional value of the petiole to a mobile device.

[0008] According to some aspects, a non-transitory computer-readable medium is provided comprising instructions executable by a processor that, when executed, may configure the processor to receive leaf spectral data generated by a portable spectrophotometer based on non-destructive spectral measurements of leaves of plants in a crop field, store the leaf spectral data in a memory, input the leaf spectral data to one or more machine learning models trained to provide an output of a nutritional value of a petiole based on the input of the leaf spectral data, receive an output comprising a nutritional value of the petiole determined by the one or more machine learning models based on the leaf spectral data, and provide a near real-time result indicative of a nutritional value of the petiole.

[0009]

[0009] According to some aspects, a system for non-destructively determining a nutritional value of a petiole in near real time from a leaf of a plant in a crop field is provided. The system may include a portable spectrophotometer and a remote server. The portable spectrophotometer may be configured to generate leaf spectral data based on a non-destructive spectral measurement of a leaf of a plant in the crop field and transmit the leaf spectral data over a network for receipt by the remote server. The remote server may include a processor and a memory, the memory storing one or more machine learning models trained to provide an output of a nutritional value of the petiole based on an input of the leaf spectral data. The remote server may be configured to receive the leaf spectral data, store the leaf spectral data in the memory, input the leaf spectral data to the one or more machine learning models, receive the nutritional value of the petiole determined by the one or more machine learning models based on the leaf spectral data, and provide a near real time result indicative of the nutritional value of the petiole to a mobile device.

[0010] According to some aspects, a method for determining refined petiole nutritional value based on leaf spectral data from a plant leaf is provided. The method may include: (a) storing the leaf spectral data in a memory; (b) converting the leaf spectral data to a preliminary petiole nutritional value by inputting the leaf spectral data using a processor to one or more linear machine learning models trained to provide an output of petiole nutritional value based on the input of the leaf spectral data, the one or more linear machine learning models outputting the preliminary petiole nutritional value; (c) refining the preliminary petiole nutritional value by inputting the preliminary petiole nutritional value using a processor to one or more non-linear machine learning models trained to provide an output of refined petiole nutritional value based on the input of the preliminary petiole nutritional value, the one or more non-linear machine learning models outputting the refined petiole nutritional value; and (d) outputting a plant nutrient report indicating the refined petiole nutritional value.

[0011]

[0011] According to some aspects, a non-transitory computer-readable medium is provided comprising instructions executable by a processor. The instructions, when executed, may configure the processor to: store leaf spectral data from leaves of a plant in a memory; convert, using the processor, the leaf spectral data into a preliminary petiole nutritional value by inputting the leaf spectral data into one or more linear machine learning models trained to provide an output of a petiole nutritional value based on an input of the leaf spectral data; receive an output comprising the preliminary petiole nutritional value from the one or more linear machine learning models; refine, using the processor, the preliminary petiole nutritional value by inputting the preliminary petiole nutritional value into one or more non-linear machine learning models trained to provide an output of a refined petiole nutritional value based on an input of the preliminary petiole nutritional value; receive an output comprising the refined petiole nutritional value from the one or more non-linear machine learning models; and output a plant nutrient report indicating the refined petiole nutritional value.

[0012]

[0012] According to some aspects, a system for determining refined petiole nutritional value based on leaf spectral data from a plant leaf is provided. The system may include a memory and a processor. The memory may be configured to store leaf spectral data from a plant leaf. The processor may be configured to convert the leaf spectral data to a preliminary petiole nutritional value by inputting the leaf spectral data to one or more linear machine learning models trained to provide an output of petiole nutritional value based on an input of the leaf spectral data, receive an output comprising the preliminary petiole nutritional value from the one or more linear machine learning models, refine the preliminary petiole nutritional value by inputting the preliminary petiole nutritional value to one or more non-linear machine learning models trained to provide an output of refined petiole nutritional value based on the input of the preliminary petiole nutritional value, receive an output comprising the refined petiole nutritional value from the one or more non-linear machine learning models, and output a plant nutrient report indicating the refined petiole nutritional value. [Brief description of the drawings]

[0013]

[0013] The drawings included herein are intended to illustrate various examples of the articles, methods, and apparatus of the present specification and are not intended to limit the scope of the teachings in any way.

[0014] [Figure 1]

[0014] FIG. 1 is a diagram of a plant leaf whose nutritional value can be determined, according to one embodiment. [Diagram 2]

[0015] FIG. 1 is a schematic diagram of a system for non-destructively determining the nutritional value of petioles from leaves of plants in a crop field in near real time, according to one embodiment. [Diagram 3]

[0016] FIG. 3 is a schematic diagram of results provided by the system of FIG. 2 and graphically displayed on a mobile device showing the nutritional value of petioles. [Figure 4]

[0017] FIG. 1 is a schematic diagram of an apparatus for non-destructively determining the nutritional value of petioles from leaves of plants in a crop field in near real time, according to one embodiment. [Diagram 5]

[0018] 1 is a flow chart illustrating an exemplary method for non-destructively determining the nutritional value of petioles from leaves of plants in a crop field in near real time using a portable spectrophotometer. [Figure 6]

[0019] 1 is a table illustrating interrelationships in nutrient uptake characteristics of potato plants according to one embodiment. [Figure 7]

[0020] 1 is a flow chart illustrating an exemplary method for determining refined petiole nutritional value based on leaf spectral data from plant leaves. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0015]

[0021] Many embodiments are described in this application, which are presented for illustrative purposes only. The described embodiments are not intended to be limiting in any sense. The present invention is broadly applicable to many embodiments, as is readily apparent from the disclosure herein. Those skilled in the art will recognize that the present invention can be implemented with modifications and variations without departing from the teachings disclosed herein. Although certain features of the present invention will be described with reference to one or more specific embodiments or figures, it should be understood that such features are not limited to use in the one or more specific embodiments or figures referenced in the description.

[0016]

[0022] The terms "an embodiment," "embodiment," "embodiments," "the embodiment," "the embodiment," "one or more embodiments," "some embodiments," and "one embodiment" mean "one or more (but not all) embodiments of the invention," unless otherwise specified.

[0017]

[0023] The terms "including," "comprising," and variations thereof mean "including, but not limited to," unless otherwise specified. A list of items does not imply that any or all of the items are mutually exclusive, unless otherwise specified. The terms "a," "an," and "the" mean "one or more," unless otherwise specified.

[0018]

[0024] As used herein and in the claims, when two or more parts are joined or cooperate directly or indirectly (i.e., through one or more intermediate parts), the parts are said to be "coupled," "connected," "attached," "joined," "fixed," or "fastened" so long as a connection occurs. As used herein and in the claims, when two or more parts are connected to one another with physical contact, the parts are said to be "directly coupled," "directly connected," "directly attached," "directly joined," "directly fixed," or "directly fastened." As used herein, when two or more parts are coupled to one another such that they move as a unit while maintaining a constant orientation relative to one another, the parts are said to be "rigidly coupled," "rigidly connected," "rigidly attached," "rigidly joined," "rigidly fixed," or "rigidly fastened." None of the terms "coupled," "connected," "attached," "joined," "fixed," and "fastened" distinguish the manner in which two or more parts are joined.

[0019]

[0025] Also, although method steps may be described (in this disclosure and / or claims) in a certain sequential order, such methods may be configured to work in alternative orders. In other words, any order or sequence of steps that may be described does not necessarily indicate a requirement that the steps be performed in that order. Steps of methods described herein may be performed in any order that is practical. Also, some steps may be performed simultaneously.

[0020]

[0026] As used in this specification and claims, a group of elements is said to "collectively" perform an action if the action is performed by any one of the elements in the group or by two or more (or all) of the elements in the group cooperatively.

[0021]

[0027] As used in this specification and claims, unless specifically stated otherwise, a first element is said to be "accepted by" a second element if at least a portion of the first element is accepted by the second element.

[0022]

[0028] In this specification, some elements may be identified by a part number, which consists of a base number followed by an alphabetic or subscript numeric suffix (e.g., 112a, or 1121). In this specification, multiple elements may be identified by part numbers that share a common base number but have different suffixes (e.g., 1121, 1122, and 1123). All elements having a common base number may be collectively or generically referred to using the base number without the suffix (e.g., 112).

[0023]

[0029] As used herein and in the claims, the terms "top," "bottom," "upper," "down," "upward," "vertical," "altitude," "upper," "lower," and similar terms generally refer to an orientation that is aligned with gravity (e.g., parallel to gravity). The terms "distal," "proximal," and similar terms generally refer to an orientation that is transverse (e.g., perpendicular) to gravity. However, none of the terms mentioned in this paragraph imply any particular alignment between elements. For example, if a first element is at a higher altitude than a second element, the first element may be said to be "vertically above" the second element, regardless of whether the first element is vertically aligned with the second element.

[0024]

[0030] Generally, plant tissue tests to determine the nutritional status of a plant are destructive and do not provide real-time results. For example, a farmer may harvest petioles from plants in a crop field for measurement. The harvested petioles will be transported to a laboratory for testing. Depending on the test being performed, a large number of petioles may need to be destructively removed.

[0025]

[0031] A laboratory can then perform plant tissue testing to provide the nutritional value of the petiole to the farmer. The process of removing, transporting, testing, and communicating the results can take several weeks to complete. This delay undermines the effectiveness of any corrective action taken based on the results (e.g., adjusting fertilizer composition and amount) because the measured plant nutrient content may have changed in the intervening period.

[0026]

[0032] The disclosed system, apparatus, and method can provide a non-destructive and near real-time determination of the nutritional status of a plant using a portable spectrophotometer. Molecules absorb light at frequencies characteristic of their structure. The portable spectrophotometer can be used to make non-destructive spectral measurements that indicate the presence and concentration of various molecules (nutrients) within a probed area (e.g., a plant leaf). The portable spectrophotometer can generate leaf spectral data (based on the non-destructive spectral measurements) that can be used to determine the nutritional status of the plant. There is no need to remove the plant leaves or petioles from the plant for testing. The disclosed system, apparatus, and method can provide the nutritional status of a plant without destroying any of the plants used for testing.

[0027]

[0033] Additionally, the disclosed systems, devices, and methods can perform the analysis and provide results in near real time, for example, within 10-20 seconds. This can allow any necessary corrective action (based on the determined nutritional status of the plant) to be implemented immediately. As used herein and in the claims, an action is said to be implemented in "near real time" if the action is implemented in less than one minute from when it is initiated.

[0028]

[0034] Canopy reflectance measurements use a terrestrial or aerial platform to determine the ratio of the amount of light leaving the canopy to the amount of light entering. Canopy reflectance measurements can be used to calculate the nutritional status of a crop. However, canopy reflectance measurements are frequently subject to atmospheric and soil interference. The disclosed systems, apparatus, and methods can probe plant leaves using a handheld spectrophotometer, avoiding the atmospheric and / or soil interference associated with making canopy reflectance measurements using a terrestrial or aerial platform.

[0029]

[0035] 1, there is shown a diagram of a plant 10 whose nutritional status may be determined using the disclosed systems, devices, or methods. In the illustrated example, the plant 10 includes a stem 14, a leaf 18, and a petiole 22 that attaches the leaf 18 to the stem 14. The leaf 18 includes a leaf blade 26 and a number of veins 30.

[0030]

[0036] The surface area of ​​the leaf blade 26 may be much greater compared to the petiole 22, and therefore, spectral reflectance measurements may be taken from the leaf blade 26 rather than the petiole 22. Thus, the measured nutritional value may be indicative of the nutritional value corresponding to the leaf blade rather than the nutritional value of the petiole.

[0031]

[0037] The nutritional value in the petiole is not the same as the nutritional value measured in other parts of the plant (e.g., the leaf blade). Currently, and historically, farmers send the petiole (specifically, the fourth petiole, counting from the upper tip of the plant) to the laboratory, and the laboratory returns the nutritional value of the petiole. Thus, farmers generally understand what corrective action is needed based on the nutritional value of the petiole in the laboratory, not based on the nutritional value associated with other parts of the plant.

[0032]

[0038] The disclosed systems, devices, and methods can take spectral measurements from the leaf blade of a plant to generate leaf spectral data and calculate the nutritional value of the petiole based on the leaf spectral data. Providing the nutritional value of the petiole (rather than the nutritional value associated with the leaf) allows a user to select and apply corrective actions that are well known and understood.

[0033]

[0039] 2, there is shown a schematic diagram of a system 100 for non-destructively determining the nutritional value of petioles from leaves 18 of plants in a crop field 34 in near real time, according to one embodiment. As shown, the system 100 may include a portable spectrophotometer 104 and a remote server 108.

[0034]

[0040] In the illustrated example, the portable spectrophotometer 104 and the remote server 108 are communicatively coupled using a network 112. The network 112 may include a communications network, such as the Internet, a wide area network (WAN), a local area network (LAN), or another type of network. In other examples, the portable spectrophotometer 104 and the remote server 108 may not be connected to a common network and may communicate using intermediate devices and / or communications networks.

[0035]

[0041] Crop field 34 may include a plurality of plants 10 (e.g., at least 500 plants). Each plant 10 may have one or more leaves 18. Plants 10 may include any suitable plant grown for agricultural purposes. For example, plants 10 may include wheat, corn, soybeans, potatoes, etc. Crop field 34 may be located outdoors or in a greenhouse.

[0036]

[0042] The portable spectrophotometer 104 may have any suitable design that is portable and capable of making non-destructive spectral measurements of the leaves of the plant 10. The portable spectrophotometer 104 may be sized and designed to be carried by an operator while the operator makes spectral measurements using the portable spectrophotometer. For example, the portable spectrophotometer 104 may be battery powered and designed to be carried in a backpack by the operator 38.

[0037]

[0043] A light source within the portable spectrophotometer 104 can generate light incident on the leaves, and a detector within the portable spectrophotometer 104 can measure the light reflected from the leaves of the plant 10. The portable spectrophotometer 104 can generate spectral data for the leaves based on the reflected spectral measurements.

[0038]

[0044] In some embodiments, the portable spectrophotometer 104 may perform spectral measurements to generate leaf spectral data in the visible and / or near infrared wavelength ranges. For example, the portable spectrophotometer 104 may generate leaf spectral data in a wavelength range of 400 nm to 2500 nm. In other examples, the portable spectrophotometer 104 may generate leaf spectral data in a smaller (e.g., 400 nm to 2000 nm) or larger (e.g., 350 nm to 2500 nm) wavelength range. A smaller wavelength range may allow for faster completion of the spectral measurements and / or require fewer resources to calculate the nutritional value of the petiole. A larger wavelength range may allow for a higher accuracy in determining the nutritional value of the petiole and / or detection of a larger number of nutrients. The wavelength range may be adjusted manually or automatically based on the nutritional value being measured. For example, the operator 38 may manually adjust the wavelength range. In other examples, the wavelength range may be adjusted automatically, for example by the remote server 108.

[0039]

[0045] The portable spectrophotometer 104 may perform spectral measurements at discrete wavelength intervals within the wavelength range being measured. For example, the portable spectrophotometer 104 may perform spectral measurements at 0.5 nm wavelength intervals within the wavelength range being measured. The wavelength intervals used in the measurements may be adjusted manually or automatically based on the nutritional value being determined. For example, the operator 38 may manually adjust the wavelength interval. In other examples, the wavelength intervals may be adjusted automatically, for example, by the remote server 108. A variety of wavelength intervals ranging from 0.1 nm to 10 nm may be used. Smaller wavelength intervals may allow for a more accurate determination of nutritional value and / or for a larger number of nutrients. Larger wavelength intervals may allow for a faster determination of nutritional value and / or require less computational resources for the determination of nutritional value.

[0040]

[0046] In some embodiments, the portable spectrophotometer 104 may be a NIRS™ DS2500 analyzer or an ASD™ FieldSpec4 spectroradiometer, while in other embodiments, the portable spectrophotometer 104 may be any other suitable spectrophotometer.

[0041]

[0047] The portable spectrophotometer 104 may be configured to transmit the generated leaf spectral data for receipt by the remote server 108. In the illustrated example, the portable spectrophotometer 104 may transmit the generated leaf spectral data to the remote server 108 over the network 112. In other examples, the portable spectrophotometer 104 may transmit the generated leaf spectral data to an intermediate device, such as the mobile device 42 of the operator 38. The portable spectrophotometer 104 may transmit the generated leaf spectral data to the mobile device 42 using point-to-point communication or a local area network (e.g., Bluetooth™), which may then transmit the leaf spectral data to the remote server 108 using a large area network / Internet.

[0042]

[0048] The mobile device 42 may be any suitable device that is portable and capable of transmitting and receiving data via a communication network (e.g., a wireless communication network such as a cellular network and / or a wi-fi network). For example, the mobile device 42 may be a laptop, tablet device, or smartphone with wireless communication capabilities. The mobile device 42 may be a dedicated companion device designed for use with the portable spectrophotometer 104. In some embodiments, the mobile device 42 may be a general-purpose device that includes proprietary software for communicating with and / or analyzing data received from the portable spectrophotometer 104. In some embodiments, the functionality of the portable spectrophotometer 104 and the mobile device 42 may be provided by a single device.

[0043]

[0049] In the illustrated example, the remote server 108 includes a processor 116 and a memory 120. The processor 116 may control the operation of the remote server 108. The processor 116 may be any suitable processor, controller, or digital signal processor capable of providing sufficient processing power depending on the configuration, purpose, and requirements of the system 100, as known to those skilled in the art. For example, the processor 116 may be a high-performance general-purpose processor. For example, the processor 116 may include a standard processor, such as an Intel® processor or an AMD® processor. Alternatively, the processor 116 may include two or more processors, each configured to perform a different dedicated task. Alternatively, dedicated hardware (e.g., a graphics processing unit (GPU)) may be used to provide some of the functionality provided by the processor 116.

[0044]

[0050] The memory 120 may include one or more of a random access memory (RAM), a read only memory (ROM), a hard drive, and a flash memory (e.g., a solid state drive). The remote server 108 may store the received leaf spectral data in the memory 120. For example, the remote server 108 may receive the leaf spectral data generated by the portable spectrophotometer 104 and store the received leaf spectral data in the memory 120.

[0045]

[0051] The memory 120 may also store a machine learning (ML) model trained to provide an output of the nutritional value of the petiole based on an input of the spectral data of the leaf. The machine learning model may be generated and trained by the remote server 108. In some embodiments, the remote server 108 may receive a machine learning model generated by an external device. The remote server 108 may train the received machine learning model and store the trained machine learning model in the memory 120. In other embodiments, the trained machine learning model may be received by the remote server 108 from an external device. The remote server 108 may store the received trained machine learning model in the memory 120.

[0046]

[0052] In various embodiments, any one or more (or all) of the memories described herein as storing one or more machine learning models (e.g., memory 120) may store a machine learning pipeline that includes the one or more machine learning models as well as associated data preprocessing code (e.g., for building, training, and / or evaluating the one or more machine learning models).

[0047]

[0053] The leaf spectral data input provided to the machine learning model may include, for example, leaf spectral data generated by the portable spectrophotometer 104 by taking spectral measurements from a plant's leaf blade. The petiole nutritional value output provided by the machine learning model may include, for example, petiole nutritional value determined for one or more nutrients, such as N, P, K, Ca, Mg, S, Mn, Zn, Fe, B, Cu, Al, and Na. The petiole nutritional values ​​for various nutrients may be expressed as percentage values ​​(of the total mass), g / kg, and / or parts per million (ppm).

[0048]

[0054] The machine learning model may be any suitable machine learning model based on the input data to be analyzed, the output data requirements, the available training data, and / or the available computational resources. In some embodiments, the machine learning model may be composed of one or more univariate or multivariate multiple linear regression (MLR) models. In the MLR models, lasso or ridge regularization may be utilized to generate lasso or ridge regression models. In other embodiments, the machine learning model may be a different model, for example, a polynomial regression model. In other embodiments, other data pre-processing steps may also be used, such as dimensionality reduction through the maximum relevance minimum redundancy (MRMR) method.

[0049]

[0055] Training data for the machine learning model may be generated by taking spectral measurements of a plant (e.g., reflectance spectral measurements from a leaf blade portion) to generate spectral data of the leaf that acts as a predictor (x) for the machine learning model. The corresponding petiole of the plant may be removed and used for laboratory histology analysis. The results of the laboratory histology analysis of the petiole may act as the response (y) for the machine learning model. A model of correlation may be constructed between the response (y) and the predictor (x). Equation (1) below represents the correlation coefficient β i 1 provides an example of an MLR model using

[0050]

number

[0051]

[0056] When the number of predictors (x) is larger than the number of responses (y), overfitting of the MLR model may occur. The accuracy of the MLR model may be improved by selecting a subset of predictors using a subset selection method. In some embodiments, Lasso MLR modeling may be used as the subset selection method. In other embodiments, other subset selection methods such as stepwise subset selection or Bayesian model averaging may be used.

[0052]

[0057] Lasso MLR modeling may be used to identify the most informative and least redundant predictors (wavelengths) using a complexity parameter (λ) that controls the amount of shrinkage. Larger values ​​of λ may result in greater penalization of non-zero coefficients in the model, resulting in greater shrinkage imposed on the coefficient values. A lasso regression model may be generated by selecting a value of λ that minimizes the root mean square error (RMSE). The chosen λ parameter may determine the number of coefficients that will make up the lasso regression model, which is selected as having the greatest explanatory power for the predicted target variable. For example, model training and performance evaluation may be performed using K-fold cross-validation using various integer values ​​of K, such as 5, and the value of λ is chosen based on the minimum RMSE. Table 1 below shows the number, range, and values ​​of the first four significant wavelength bands resulting from lasso MLR modeling of various elements for example leaf spectral data obtained from potato plants.

[0053] [Table 1]

[0054]

[0058] The processor 116 may input the received (e.g., from the portable spectrophotometer 104) and / or stored (e.g., in the memory 120) leaf spectral data into the trained machine learning model. The machine learning model may estimate the nutritional value of the petiole based on the leaf spectral data. For example, the machine learning model may estimate the nutritional value of the petiole for N, P, K, Ca, Mg, S, Mn, Zn, Fe, B, Cu, Al, and Na based on the leaf spectral data generated by taking spectral measurements of the leaf blades of potato plants. The processor 116 may receive the nutritional value of the petiole determined by the machine learning model.

[0055]

[0059] The processor 116 can provide near real-time results indicative of the nutritional value of the petioles to the mobile device. For example, the processor 116 may provide the near real-time results to the mobile device 42 of the operator 38. For example, the processor 116 may provide the near real-time results to the mobile device 42 in less than one minute (e.g., within 0.01 to 30 seconds) from when the portable spectrophotometer transmits the spectral measurement. This may enable the operator 38 to non-destructively collect near real-time results indicative of the nutritional value of the petioles from multiple plants in a crop field.

[0056]

[0060] Reference is now made to Figures 2 and 3. Figure 3 shows a schematic diagram of results provided by the system 100 and graphically displayed on the mobile device 42 showing the nutritional value of the petiole. The mobile device 42 may include a touch screen display 46 that provides an interactive display to the operator 38. The touch screen display 46 may provide multiple display portions 50a-50c. For example, the display portion 50a may provide a graphical display of the determined nutritional value of the petiole. In the illustrated example, the display portion 50a provides a bar graph displaying the determined nutritional value of the petiole for six nutrients. In other embodiments, the display 46 may not be configured to detect touch input.

[0057]

[0061] Display portion 50b may provide additional information regarding the determined nutritional value of the petiole, such as the number of spectral measurements of the leaf made, a confidence level associated with the determined nutritional value of the petiole, etc. Display portion 50c may provide recommended corrective actions based on the determined nutritional value of the petiole. In some examples, one or more of display portions 50 may include other information unrelated to the results indicating the nutritional value of the petiole.

[0058]

[0062] Referring now to FIG. 4, there is shown a schematic diagram of an apparatus 200 for non-destructively determining the nutritional value of petioles from leaves of plants in a crop field in near real time. In some embodiments, the apparatus 200 may be implemented as a server device (e.g., the remote server 108 shown in FIG. 2). In other embodiments, the apparatus 200 may be implemented using any other suitable hardware components. For example, the apparatus 200 may be implemented as a mobile device 42 (FIGS. 2 and 3), integrated with a portable spectrophotometer 104 (FIG. 2), or implemented as a companion device usable with a portable spectrophotometer 104 (FIG. 2).

[0059]

[0063] 4, the apparatus 200 includes a memory 208, an application 212, an output device 216, a display device 220, a secondary storage device 224, a processor 228, an input device 232, and a communication device 236. One or more (or all) of the memory 208, the application 212, the output device 216, the display device 220, the secondary storage device 224, the processor 228, the input device 232, and the communication device 236 may be communicatively coupled via wires and / or wirelessly.

[0060]

[0064] In some embodiments, the device 200 includes a plurality of any one or more of the memory 208, the application 212, the output device 216, the display device 220, the secondary storage device 224, the processor 228, the input device 232, and the communication device 236. In some embodiments, the device 200 does not include any one or more of the application 212, the secondary storage device 224, a network connection, the input device 232, the output device 216, the display device 220, and the communication device 236.

[0061]

[0065] In at least one embodiment, the device 200 includes a connection to a network 204, such as a wired or wireless connection to the Internet or to a private network. In some cases, the network 204 includes other types of computer or telecommunications networks. The device 200 may receive, via the network 204, leaf spectral data generated by a portable spectrophotometer (e.g., portable spectrophotometer 104 of FIG. 2) based on non-destructive spectral measurements of leaves of plants (e.g., potato plants) in a crop field (e.g., crop field 34 of FIG. 2).

[0062]

[0066] In some embodiments, the apparatus 200 may receive the leaf spectral data directly from the portable spectrophotometer, either wired or wirelessly, while in other embodiments, the apparatus 200 may receive the leaf spectral data generated by the portable spectrophotometer from an intermediate device (e.g., the mobile device 42 of FIG. 2).

[0063]

[0067] The memory 208 may include one or more of a random access memory (RAM) and a read only memory (ROM). In some embodiments, the memory 208 stores one or more applications 212 executed by the processor 228. The applications 212 correspond to software modules that include computer-executable instructions for performing operations for the functions and methods described herein.

[0064]

[0068] The memory 208 may store a machine learning model trained to provide an output of the nutritional value of the petiole based on an input of the spectral data of the leaf. The machine learning model may be generated and trained by the device 200. In some embodiments, the device 200 may receive the generated machine learning model from an external device (e.g., via the network 204) and train the received machine learning model. In some embodiments, the device 200 may receive the trained machine learning model from an external device (e.g., via the network 204) and store the trained machine learning model in the memory 208. The machine learning model may be any suitable model, for example, a Lasso regression model.

[0065]

[0069] In some embodiments, the memory 208 may also store the received leaf spectral data, the determined petiole nutritional value, and / or results derived from the petiole nutritional value (e.g., a graphical representation of the petiole nutritional value). The stored data may be retrieved by a user for historical data analysis.

[0066]

[0070] Secondary storage device 224 may include any suitable non-transitory computer-readable medium that contains instructions executable by a processor, such as processor 228. For example, secondary storage device 224 may include a hard drive, floppy drive, CD drive, DVD drive, Blu-ray drive, solid-state drive, flash memory, or other type of non-volatile data storage device. Processor 228 may execute instructions contained on secondary storage device 224 to perform processing for the functions and methods described herein.

[0067]

[0071] In some embodiments, apparatus 200 stores information in a remote storage device, such as cloud storage, accessible over a network, such as network 204 or another network. In some embodiments, apparatus 200 stores information in a distributed manner across multiple storage devices, such as memory 208 and secondary storage device 224 (i.e., each of the multiple storage devices stores a portion of the information and the multiple storage devices collectively store all of the information). Thus, as used herein and in the claims, storing data in a storage device can mean storing the data in a local storage device, storing the data in a remote storage device, or storing the data in a distributed manner across multiple storage devices, each of which can be local or remote.

[0068]

[0072] The input devices 232 may include any device for inputting information into the apparatus 200. For example, the input devices 232 may be a keyboard, a keypad, a cursor device, a touch screen, a camera, or a microphone. The input devices 232 may also include input ports and wireless radios (e.g., Bluetooth or 802.11x) for wired and wireless connections to external devices.

[0069]

[0073] Display device 220 may include any type of device for presenting visual information. For example, display device 220 may be a computer monitor, a flat screen display, a projector, or a display panel.

[0070]

[0074] Output device(s) 216 may include any type of device for presenting a hard copy of information, such as, for example, a printer. Output device(s) 216 may also include other types of output devices, such as, for example, speakers. In at least one embodiment, output device(s) 216 includes one or more of an output port and a radio (e.g., Bluetooth or 802.11x) for wired and wireless connections to external devices.

[0071]

[0075] The communications device 236 may have any design suitable for receiving analog and / or digital inputs and providing analog and / or digital outputs, in some embodiments, the communications device 236 may include separate modules for analog and digital signals.

[0072]

[0076] Processor 228 may be any device capable of executing applications, computer readable instructions, or programs. The applications, computer readable instructions, or programs may be stored in memory 208 or in secondary storage device 224, or may be received from a remote storage device accessible, for example, over network 204. Processor 228 may be a high performance general purpose processor, a standard processor (e.g., an Intel® processor or AMD® processor), dedicated hardware (e.g., a GPU), or multiple processing devices that collectively perform the functions provided by processor 228.

[0073]

[0077] The processor 228 may input the leaf spectral data (e.g., the leaf spectral data stored in the memory 208) to a trained machine learning model (e.g., the trained machine learning model stored in the memory 208). The machine learning model may determine the nutritional value of the petiole based on the leaf spectral data. The processor 228 may receive an output from the machine learning model including the nutritional value of the petiole determined by the machine learning model.

[0074]

[0078] In some embodiments, the processor 228 may provide near real-time results indicative of the nutritional value of the petiole to a mobile device (e.g., mobile device 42 of FIG. 2). The nutritional value of the petiole may be displayed graphically on the mobile device. In other embodiments, the processor 228 may provide results to the display device 220, and the nutritional value of the petiole may be displayed graphically on the display device 220.

[0075]

[0079] FIG. 4 illustrates an exemplary hardware overview of device 200. In alternative embodiments, device 200 includes fewer, additional, or different components. In addition, while aspects of one implementation of device 200 are described as being stored in memory, those skilled in the art will appreciate that these aspects may also be stored in or read from other types of computer program products or computer readable media, such as secondary storage devices including hard disks, floppy disks, CDs, or DVDs, carrier waves from the Internet or other networks, or other forms of RAM or ROM. For example, device 200 may include a non-transitory computer readable medium that stores computer readable instructions that, when executed by processor 228, configure processor 228 to perform one or more methods described herein.

[0076]

[0080]

[0023] Referring now to Figure 5, there is shown a flow chart illustrating an exemplary method 300 for non-destructively determining the nutritional value of petioles from leaves of plants in a crop field in near real time using a portable spectrophotometer. Method 300 may be implemented, for example, using system 100 (Figure 2) or device 200 (Figure 4), also referenced below in Figures 2 and 4.

[0077]

[0081] The method 300 may be performed while carrying a portable spectrophotometer in a crop field. For example, the portable spectrophotometer 104 may be carried by a person or vehicle in the crop field 34.

[0078]

[0082] At 308, non-destructive spectral measurements of leaves of plants in the crop field may be made using a portable spectrophotometer to generate spectral data for the leaves. For example, non-destructive spectral measurements of leaves of potato plants may be made using the portable spectrophotometer 104. The portable spectrophotometer 104 may generate the spectral data for the leaves based on the spectral measurements.

[0079]

[0083] At 312, the leaf spectral data may be stored in memory. In some embodiments, the generated leaf spectral data may be transmitted over a network to a remote server and stored in the memory of the remote server. For example, the portable spectrophotometer 104 may transmit the generated leaf spectral data over the network 112 to the remote server 108, which may store the leaf spectral data in memory 120.

[0080]

[0084] In other embodiments, the generated leaf spectral data may be transmitted to the mobile device, and the mobile device may transmit the leaf spectral data to a remote server over a network. The leaf spectral data may be stored in a memory of the remote server. For example, the portable spectrophotometer 104 may transmit the generated leaf spectral data to the mobile device 42, and the mobile device 42 may transmit the leaf spectral data to the remote server 108 over the network 112. The remote server 108 may store the leaf spectral data in memory 120.

[0081]

[0085] At 316, the processor may calculate the nutritional value of the petiole based on the stored spectral data of the leaf. The processor may calculate the nutritional value of the petiole by inputting the spectral data of the leaf into a machine learning model trained to provide an output of nutritional value of the petiole based on an input of the spectral data of the leaf. In response, the machine learning model may output the nutritional value of the petiole to the processor. For example, the processor 116 (or the processor 228) may input the spectral data of the leaf into a trained machine learning model stored in the memory 120 (or the memory 208). The machine learning model may be a Lasso regression model.

[0082]

[0086] At 320, near real-time results indicative of the nutritional value of the petiole may be provided. In some embodiments, the near real-time results may be provided by transmitting the nutritional value of the petiole to a mobile device. The nutritional value of the petiole may then be displayed graphically on the mobile device. For example, the remote server 108 or the apparatus 200 may provide the near real-time results by transmitting the nutritional value of the petiole to the mobile device 42. The mobile device 42 may display the nutritional value of the petiole graphically.

[0083]

[0087] In some embodiments, method 300 may proceed to 308 if there are additional leaves to be measured. For example, method 300 may loop from 308 to 320 for at least five other plant leaves in the crop field within 10 minutes. This may allow for near real-time results to be obtained that indicate the nutritional value of the petioles corresponding to non-destructive spectral measurements from multiple leaves in a short period of time. This, in turn, would allow the farm to simultaneously respond to the returning petiole nutritional value (e.g., by adjusting fertilizer composition and / or amount) before the measured plant nutrient content has time to change. This may improve the effectiveness of corrective actions, which may improve crop yield and / or volume.

[0084]

[0088] In other embodiments, the machine learning processing of the leaf spectral data may follow a two-tiered approach to provide a more accurate estimate of the nutritional value of the petiole (referred to herein as "refined petiole nutritional value"). The first ML tier may include one or more first machine learning models (e.g., one or more independent instances of the machine learning models described above with reference to Figures 1-5). The second ML tier may include one or more second machine learning models. The petiole nutritional value determined by the one or more first machine learning models of the first tier may be refined to create a statistically more accurate value (i.e., a value of one or more petiole nutrients that is closer to the actual value of one or more nutrients of the petiole that would have been obtained if the petiole had been removed according to conventional methods and destructively measured in a laboratory). In these embodiments, the petiole nutritional value determined by the one or more machine learning models of the first tier may be referred to as a preliminary petiole nutritional value.

[0085]

[0089] The one or more machine learning models in the first layer may be trained and stored in the manner described above. In some embodiments, only a small subset of the most explanatory features (i.e., intensity values ​​related to a particular wavelength or wavelength interval of the collected leaf spectral data) are input to the one or more first machine learning models. This addresses the problem of the dataset containing a large number of features (i.e., high dimensionality of the feature space). Before feeding these features to the one or more machine learning models in the first layer, a feature reduction step (e.g., minimum redundancy maximum relevance (MRMR) feature selection) may be performed independently for each target variable to extract a subset of features to be input to the one or more machine learning models in the first layer. Preselecting the features may improve the estimation performance of most nutrients. A model regularization step may also be performed, such as the Lasso regression model described above, which removes any features that are not considered to significantly contribute to the predictive ability of the model.

[0086]

[0090] The memory 120 (or memory 208) may store one or more machine learning models of the second tier trained to provide refined petiole nutritional value output based on the preliminary petiole nutritional value output by the one or more machine learning models of the first tier. The one or more machine learning models of the second tier may be generated and trained by the remote server 108. In some embodiments, the remote server 108 may receive the one or more machine learning models of the second tier generated by an external device. The remote server 108 may train the received one or more machine learning models of the second tier and store the trained one or more machine learning models of the second tier in the memory 120 (or memory 208). In other embodiments, the trained one or more machine learning models of the second tier may be received by the remote server 108 from an external device. The remote server 108 may store the received trained one or more machine learning models of the second tier in the memory 120 (or memory 208).

[0087]

[0091] The one or more machine learning models of the second tier may be configured with any suitable type of machine learning model based on the input data to be analyzed, the output data requirements, the available training data, and / or the available computational resources. In some embodiments, one or more (or all) of the one or more second machine learning models used in the second tier may be a nonlinear multiple regression model, such as, but not limited to, Support Vector Regression (SVR), Random Forest (RF), or Extreme Gradient Boosting (XGB).

[0088]

[0092] As described above, the first tier of one or more first machine learning models may be trained using a set of data that includes a matched set of leaf spectral data and laboratory measured nutritional value of petioles, which allows the first tier of one or more machine learning models to establish a relationship between the leaf spectral data values ​​and the nutritional value of individual petioles.

[0089]

[0093] In general, there are overlapping effects at different wavelengths on reflectance due to the complex relationship between nutrient concentrations in plants. A species of plant will have a tendency to take up various nutrients (i.e., the species has a specific nutrient uptake characteristic). The uptake of a nutrient will be influenced by the uptake of other nutrients. For example, for a species of plant, when the plant takes up a certain nutrient "x", it also tends to take up nutrient "y" (i.e., there is a positive correlation between nutrient "x" and nutrient "y"). However, when the plant takes up nutrient "x", it also tends to reduce the uptake of nutrient "z" (i.e., there is a negative correlation between nutrient "x" and nutrient "z"). For example, potassium (K) and magnesium (Mg) are known for their antagonistic action as competing cations for plant uptake.

[0090]

[0094] Therefore, refining the preliminary petiole nutritional value determined from leaf spectral data based on the inter-linkages between nutrients for a given plant species may provide more statistically accurate results.

[0091]

[0095] The nonlinearity of the second layer one or more second machine learning models allows the model or models to simultaneously consider these nutrient uptake characteristics and potentially correlated outcomes. The second layer one or more second machine learning models can be trained using a set of data that includes the nutritional value of a large number of different petioles. The second layer one or more second machine learning models can process this data to establish the uptake relationship between various nutrients for a given plant species. Importantly, the nutrient uptake characteristics of different plant species will be different.

[0092]

[0096] FIG. 6 shows a table 400 demonstrating an example of nutrient uptake characteristics of a potato plant species. As seen in FIG. 6, for example, potato plant nitrogen (N) uptake tends to promote phosphorus (P) uptake (indicated by a positive value of 0.83). However, potato plants also tend to reduce boron (B) uptake when uptakes nitrogen (indicated by a negative value of −0.66). In other words, there is a positive correlation between nitrogen uptake and phosphorus uptake, but a negative correlation between nitrogen uptake and boron uptake. As shown in FIG. 6, correlations between each predicted petiole nutrient can be determined. Considering the correlations between the uptakes of various nutrients allows the refined petiole nutritional value to be statistically more accurate than the preliminary petiole nutritional value output from one or more first machine learning models of the first layer to which the leaf spectral data is input.

[0093]

[0097] The reasons for positive and negative correlations in the uptake of various nutrients are manifold. The following describes some examples that may apply to one or more species of potato plants. A positive relationship between N and P may be because N fixation may depend on P availability in the soil for plant uptake. An antagonistic effect between calcium (Ca) and P may be caused by the precipitation of less soluble calcium phosphate in the vicinity of the nutrient-absorbing roots. K and Mg may be competing cations for plant uptake. Ca and Mg may have a synergistic contribution to maintaining a stable pH in the soil. N may be negatively coupled with Mg, as an increase in the rate of N may be reduced by the addition of a rate of Mg. B is negatively correlated with P, but the availability of B is positively affected by the presence of Ca. Plant uptake of K may be attributed to N for its role in leaf area development and photosynthesis. The form of K application as sulfate may lead to assimilation partitioning of sulfur (S). For example, some K fertilizers contain potassium as K2SO4, and therefore potassium application is accompanied by sulfur. An increase in K supply (e.g. to the ground via fertilizer) typically results in an increase in K uptake by the plant, which may decrease manganese (Mn) concentrations in the leaves. An increase in S supply (e.g. to the ground via fertilizer) typically results in an increase in S uptake by the plant, which may increase K and P uptake. High doses of K may decrease iron (Fe) uptake due to improved resistance to Fe toxicity. Translocation of Ca from roots to leaves may be disrupted in the presence of Cu, so there is an opposition between Cu and Ca.

[0094]

[0098]

[0041] Referring now to Figure 7, there is shown a flow chart illustrating an exemplary method 500 for determining the nutritional value of refined petioles based on leaf spectral data from plant leaves. Method 500 may be implemented, for example, using system 100 (Figure 2) or device 200 (Figure 4), also referred to below in Figures 2 and 4.

[0095]

[0099] Method 500 may be performed while carrying a portable spectrophotometer in the crop field. For example, portable spectrophotometer 104 may be carried by a person or vehicle in the crop field 34. Method 500 may also be performed using any other suitable spectrophotometer, including non-portable spectrophotometers.

[0096]

[0100] Method 500 may optionally include step 504. At 504, a spectral measurement may be taken of a plant leaf using a spectrophotometer (e.g., the portable spectrophotometer 104) to generate spectral data for the leaf. For example, a spectral measurement may be taken of a potato plant leaf.

[0097]

[0101] At 506, the leaf spectral data may be stored in memory. In some embodiments, the generated leaf spectral data may be transmitted over a network to a remote server and stored in the memory of the remote server. For example, the portable spectrophotometer 104 may transmit the generated leaf spectral data over the network 112 to the remote server 108, which may store the leaf spectral data in memory 120.

[0098]

[0102] In other embodiments, the generated leaf spectral data may be transmitted to the mobile device, and the mobile device may transmit the leaf spectral data to a remote server over a network. The leaf spectral data may be stored in a memory of the remote server. For example, the portable spectrophotometer 104 may transmit the generated leaf spectral data to the mobile device 42, and the mobile device 42 may transmit the leaf spectral data to the remote server 108 over the network 112. The remote server 108 may store the leaf spectral data in memory 120.

[0099]

[0103] At 508, the processor may convert the leaf spectral data into a preliminary petiole nutritional value. The processor may convert the leaf spectral data by inputting the leaf spectral data into one or more linear machine learning models trained to provide an output of petiole nutritional value based on an input of the leaf spectral data. In response, the one or more linear machine learning models may output the preliminary petiole nutritional value to the processor. For example, the processor 116 (or the processor 228) may input the leaf spectral data into one or more trained linear machine learning models stored in the memory 120 (or the memory 208). One or more (or all) of the one or more linear machine learning models may be a Lasso regression model.

[0100]

[0104] At 510, the processor may refine the preliminary petiole nutritional value by inputting the preliminary petiole nutritional value into one or more nonlinear machine learning models trained to provide an output of refined petiole nutritional value based on the input of the preliminary petiole nutritional value. In response, the one or more nonlinear machine learning models may output the refined petiole nutritional value to the processor. For example, the processor 116 (or the processor 228) may input the preliminary petiole nutritional value into one or more trained nonlinear machine learning models stored in the memory 120 (or the memory 208). One or more (or all) of the one or more nonlinear machine learning models may be multivariate regression models.

[0101]

[0105] At 512, a phytonutrient report may be output showing the nutritional value of the refined petiole. The phytonutrient report may have any suitable format, including but not limited to a list, table, chart, graphical representation, any type of visualization (e.g., infographic, map, etc.), or animation of the nutritional value of the refined petiole. In some embodiments, the phytonutrient report may be output by transmitting the nutritional value of the refined petiole to a mobile device. For example, the remote server 108 or the apparatus 200 may provide near real-time results (e.g., less than one minute from the completion of step 506) by transmitting the nutritional value of the refined petiole to the mobile device 42. The mobile device 42 may graphically display the nutritional value of the petiole.

[0102]

[0106] In some embodiments, method 500 may proceed to steps 504 or 506 if there are additional leaves to be measured. For example, method 500 may loop from 504 and / or 506 to 512 for at least five other plant leaves in the crop field within approximately one hour. This may allow results to be obtained that indicate refined petiole nutritional value corresponding to spectral measurements from multiple leaves in a short period of time. This in turn would allow the farm to simultaneously react to the refined petiole nutritional value returned (e.g., by adjusting fertilizer composition and / or amount) before the measured plant nutrient content has time to change. This may improve the effectiveness of corrective measures, which may improve crop yield and / or volume.

[0103]

[0107] At 514, a treatment may optionally be applied to the crop (e.g., including and / or proximate to the measured plant). The treatment may include applying (e.g., spraying or scattering) liquid and / or solid (e.g., granular) fertilizer to the crop. The treatment may be determined (e.g., by processor 116 or 228) based at least in part on the refined nutrient values ​​of step 512. The treatment may be applied under the control of a human operator and / or as directed by electronic automation (e.g., as directed by processor 116 or 228).

[0104]

[0108] The two-tier approach described herein, which includes linear and nonlinear machine learning models, can substantially increase the accuracy of nutrient prediction. In particular, the estimation of Zn, Fe, and Al is substantially more accurate than using only one or more linear machine learning models in a one-tier approach.

[0105]

[0109] Although the above description presents examples of embodiments, it will be understood that some features and / or functions of the described embodiments can be modified without departing from the spirit and operating principles of the described embodiments. Therefore, what has been described above is intended to be illustrative and non-limiting of the present invention, and those skilled in the art will understand that other variations and modifications can be made without departing from the scope of the present invention defined in the claims appended hereto. The claims should not be limited by the preferred embodiments and examples, but should be given the broadest interpretation consistent with the entire specification.

[0106] item Item 1: A method for non-destructively determining in near real time a nutritional value of a petiole from leaves of plants in a crop field using a portable spectrophotometer, the method comprising: (a) performing non-destructive spectral measurements of leaves of plants in a crop field using the portable spectrophotometer to generate leaf spectral data; (b) storing the leaf spectral data in a memory; (c) calculating, by a processor, a nutritional value of the petiole based on the stored leaf spectral data; and (d) providing a near real time result indicative of the nutritional value of the petiole. Item 2: The method of item 1, wherein the providing comprises transmitting the nutritional value of the petiole to a mobile device. Item 3: The method of items 1 or 2, wherein the providing step further comprises graphically displaying the nutritional value of the petiole on a mobile device. Item 4: The method of items 1 to 3, wherein the storing step comprises transmitting the leaf spectral data to a remote server via a network, the remote server comprising a memory and a processor. Item 5: The method of items 1 to 4, wherein the storing step comprises the mobile device receiving the leaf spectral data and the mobile device transmitting the leaf spectral data to a remote server via a network, the remote server comprising a memory and a processor. Item 6: The method of items 1 to 5, wherein the providing comprises transmitting the nutritional value of the petiole from a remote server to the mobile device and graphically displaying the nutritional value of the petiole on the mobile device. Item 7: The method of items 1 to 6, further comprising repeating steps (a) to (d) for at least five other plant leaves in the crop field within 10 minutes. Item 8: The method of items 1 to 7, wherein the plant is a potato plant. Item 9: The method of items 1 to 8, wherein the calculating step comprises inputting the spectral data of the leaf to one or more machine learning models trained to provide an output of nutritional value of the petiole based on an input of the spectral data of the leaf, wherein the one or more machine learning models output nutritional value of the petiole. Item 10: The method of items 1 to 9, wherein each of the one or more machine learning models is a Lasso regression model. Item 11: An apparatus for non-destructively determining nutritional value of petioles from leaves of plants in a crop field in near real time, the apparatus comprising: a memory storing one or more machine learning models trained to provide an output of nutritional value of the petioles based on an input of leaf spectral data; and at least one processor communicatively coupled to the memory, wherein the at least one processor is collectively configured to receive leaf spectral data generated by a portable spectrophotometer based on non-destructive spectral measurements of leaves of plants in the crop field, store the leaf spectral data in the memory, input the leaf spectral data to the one or more machine learning models, receive an output comprising nutritional value of the petioles determined by the one or more machine learning models based on the leaf spectral data, and provide a near real-time result indicative of nutritional value of the petioles to a mobile device. Item 12: The device of Item 11, in which the nutritional value of the petiole is displayed graphically on a mobile device. Item 13: The apparatus of item 11 or 12, wherein the leaf spectral data generated by the portable spectrophotometer is received from a mobile device via a network. Item 14: The apparatus of items 11 to 13, wherein the plant is a potato plant. Item 15: The apparatus of items 11 to 14, wherein each of the one or more machine learning models is a Lasso regression model. Item 16: The device of items 11 to 15, wherein the memory comprises one or more of a random access memory (RAM), a read-only memory (ROM), a hard drive, and a flash memory. Item 17: A non-transitory computer-readable medium comprising instructions executable by a processor, the instructions, when executed, configure the processor to receive leaf spectral data generated by a portable spectrophotometer based on non-destructive spectral measurements of leaves of plants in a crop field, store the leaf spectral data in a memory, input the leaf spectral data to one or more machine learning models trained to provide an output of nutritional value of the petiole based on the input of the leaf spectral data, receive an output comprising a nutritional value of the petiole determined by the one or more machine learning models based on the leaf spectral data, and provide a near real-time result indicative of the nutritional value of the petiole. Item 18: The non-transitory computer-readable medium of item 17, wherein the leaf spectral data generated by the portable spectrophotometer is received from a mobile device via a network. Item 19: The non-transitory computer-readable medium of items 17 or 18, wherein the plant is a potato plant. Item 20: The non-transitory computer-readable medium of items 17 to 19, wherein each of the one or more machine learning models is a Lasso regression model. Item 21: The non-transitory computer-readable medium of items 17 to 20, wherein the memory comprises one or more of a random access memory (RAM), a read-only memory (ROM), a hard drive, and a flash memory. Item 22: A system for non-destructively determining nutritional value of petioles in near real time from leaves of plants in a crop field, the system comprising: a portable spectrophotometer configured to generate leaf spectral data based on non-destructive spectral measurements of leaves of plants in the crop field and transmit the leaf spectral data over a network for receipt by a remote server, the remote server comprising a processor and a memory, the memory storing one or more machine learning models trained to provide an output of nutritional value of the petioles based on an input of the leaf spectral data, the remote server configured to receive the leaf spectral data, store the leaf spectral data in the memory, input the leaf spectral data to the one or more machine learning models, receive the nutritional value of the petioles determined by the one or more machine learning models based on the leaf spectral data, and provide a near real-time result indicative of the nutritional value of the petioles to a mobile device. Item 23: The system of Item 22, in which the nutritional value of the petiole is displayed graphically on a mobile device. Item 24: The system of items 22 or 23, wherein the portable spectrophotometer is configured to transmit the leaf spectral data to a mobile device, and the mobile device transmits the leaf spectral data to a remote server. Item 25: The system of items 22 to 24, in which the plant is a potato plant. Item 26: The system of items 22 to 25, wherein each of the one or more machine learning models is a Lasso regression model. Item 27: The system of items 22 to 26, wherein the memory comprises one or more of random access memory (RAM), read-only memory (ROM), a hard drive, and flash memory. Item 28: A method for determining a refined petiole nutritional value based on leaf spectral data from leaves of a plant, the method comprising: (a) storing the leaf spectral data in a memory; (b) converting, using a processor, the leaf spectral data to a preliminary petiole nutritional value by inputting the leaf spectral data to one or more linear machine learning models trained to provide an output of petiole nutritional value based on an input of the leaf spectral data, wherein the one or more linear machine learning models output the preliminary petiole nutritional value; (c) refining, using a processor, the preliminary petiole nutritional value by inputting the preliminary petiole nutritional value to one or more non-linear machine learning models trained to provide an output of refined petiole nutritional value based on the input of the preliminary petiole nutritional value, wherein the one or more non-linear machine learning models output the refined petiole nutritional value; and (d) outputting a plant nutrient report indicating the refined petiole nutritional value. Item 29: The method of item 28, wherein the one or more nonlinear machine learning models are trained to model interrelationships in nutrient uptake profiles of different plant nutrients. Item 30: The method of items 28 or 29, further comprising, prior to said storing, performing a spectral measurement of a plant leaf using a spectrophotometer to generate leaf spectral data. Item 31: The method of items 28 to 30, wherein each of the one or more linear machine learning models is a Lasso regression model. Item 32: The method of items 28 to 31, wherein each of the one or more nonlinear machine learning models is a multivariate regression model. Item 33: The method of items 28 to 32, wherein each of the one or more nonlinear machine learning models is one of a support vector regression (SVR) model, a random forest (RF) model, and an extreme gradient boosting (XGB) model. Item 34: The method of items 28 to 33, wherein the outputting comprises transmitting the refined nutritional value of the petiole to a mobile device. Item 35: The method of items 28 to 34, wherein the outputting step further comprises graphically displaying the nutritional value of the refined petiole on a mobile device. Item 36: The method of any one of items 28 to 35, wherein the storing step comprises transmitting the leaf spectral data via a network to a remote server, the remote server comprising a memory and a processor. Item 37: The method of any one of items 28 to 36, wherein the storing step comprises the mobile device receiving the leaf spectral data and the mobile device transmitting the leaf spectral data to a remote server via a network, the remote server comprising a memory and a processor. Item 38: The method of items 28 to 37, wherein the outputting step comprises transmitting the refined petiole nutritional value from a remote server to the mobile device and graphically displaying the refined petiole nutritional value on the mobile device. Item 39: The method of items 28 to 38, wherein the plant leaves are leaves of a potato plant. Item 40: The method of items 28 to 39, further comprising applying a fertilizer treatment to the plant based at least in part on the nutritional value of the refined petiole. Item 41: A non-transitory computer-readable medium comprising instructions executable by a processor, the instructions, when executed, configuring the processor to store leaf spectral data from leaves of the plant in a memory, convert, using the processor, the leaf spectral data to a preliminary petiole nutritional value by inputting the leaf spectral data to one or more linear machine learning models trained to provide an output of petiole nutritional value based on an input of the leaf spectral data, receive an output comprising the preliminary petiole nutritional value from the one or more linear machine learning models, refine, using the processor, the preliminary petiole nutritional value by inputting the preliminary petiole nutritional value to one or more non-linear machine learning models trained to provide an output of refined petiole nutritional value based on the input of the preliminary petiole nutritional value, receive an output comprising the refined petiole nutritional value from the one or more non-linear machine learning models, and output a plant nutrient report indicating the refined petiole nutritional value. Item 42: The non-transitory computer-readable medium of item 41, wherein the one or more non-linear machine learning models are trained to model interrelationships in nutrient uptake profiles of different plant nutrients. Item 43: The non-transitory computer-readable medium of items 41 or 42, wherein the instructions, when executed, configure the processor to receive spectral data taken from a plant leaf using a spectrophotometer. Item 44: The non-transitory computer-readable medium of items 41 to 43, wherein each of the one or more linear machine learning models is a Lasso regression model. Item 45: The non-transitory computer-readable medium of items 41 to 44, wherein each of the one or more nonlinear machine learning models is a multivariate regression model. Item 46: The non-transitory computer-readable medium of items 41 to 45, wherein each of the one or more nonlinear machine learning models is one of a support vector regression (SVR) model, a random forest (RF) model, and an extreme gradient boosting (XGB) model. Item 47: The non-transitory computer-readable medium of items 41 to 46, wherein the outputting comprises transmitting the refined nutritional value of the petiole to a mobile device. Item 48: The non-transitory computer-readable medium of items 41 to 47, wherein the instructions, when executed, configure the processor to apply a fertilizer treatment to the plant based at least in part on the nutritional value of the refined petiole. Item 49: A system for determining a refined petiole nutritional value based on leaf spectral data from leaves of a plant, comprising: a memory configured to store leaf spectral data from leaves of the plant; and a processor configured to convert the leaf spectral data to a preliminary petiole nutritional value by inputting the leaf spectral data to one or more linear machine learning models trained to provide an output of petiole nutritional value based on an input of the leaf spectral data, receive an output comprising the preliminary petiole nutritional value from the one or more linear machine learning models, refine the preliminary petiole nutritional value by inputting the preliminary petiole nutritional value to one or more non-linear machine learning models trained to provide an output of refined petiole nutritional value based on the input of the preliminary petiole nutritional value, receive an output comprising the refined petiole nutritional value from the one or more non-linear machine learning models, and output a plant nutrient report indicating the refined petiole nutritional value. Item 50: The system of item 49, wherein the one or more nonlinear machine learning models are trained to model interrelationships in nutrient uptake profiles of different plant nutrients. Item 51: The system of items 49 or 50, wherein the processor is further configured to receive spectral data taken from a plant leaf using a spectrophotometer. Item 52: The system of items 49 to 51, wherein each of the one or more linear machine learning models is a Lasso regression model. Item 53: The system of items 49 to 52, wherein each of the one or more nonlinear machine learning models is a multivariate regression model. Item 54: The system of items 49 to 53, wherein each of the one or more nonlinear machine learning models is one of a support vector regression (SVR) model, a random forest (RF) model, and an extreme gradient boosting (XGB) model. Item 55: The system of items 49 to 54, wherein the outputting step comprises transmitting the refined nutritional value of the petiole to a mobile device. Item 56: The system of items 49 to 55, wherein the processor is further configured to apply a fertilizer treatment to the plant based at least in part on the nutritional value of the refined petiole.

Claims

1. 1. A method for determining the nutritional value of a petiole refined based on spectral data of the leaf from a plant leaf, comprising: (a) storing the spectral data of the leaf in a memory; (b) using a processor, converting the leaf spectral data into a preliminary petiole nutritional value by inputting the leaf spectral data into one or more linear machine learning models trained to provide an output of a petiole nutritional value based on an input of the leaf spectral data, the one or more linear machine learning models outputting the preliminary petiole nutritional value; (c) using the processor to refine the preliminary petiole nutritional value by inputting the preliminary petiole nutritional value into one or more non-linear machine learning models trained to provide an output of refined petiole nutritional value based on an input of the preliminary petiole nutritional value, wherein the one or more non-linear machine learning models output a refined petiole nutritional value; (d) outputting a plant nutrient report indicating the nutritional value of the refined petiole; and A method comprising:

2. 2. The method of claim 1, wherein the one or more nonlinear machine learning models are trained to model interrelationships in the nutrient uptake profile of different plant nutrients.

3. 10. The method of claim 1, further comprising, prior to said storing, performing a spectral measurement of a plant leaf using a spectrophotometer to generate leaf spectral data.

4. The method of claim 1 , wherein each of the one or more nonlinear machine learning models is a multivariate regression model.

5. 2. The method of claim 1, wherein each of the one or more nonlinear machine learning models is one of a support vector regression (SVR) model, a random forest (RF) model, and an extreme gradient boosting (XGB) model.

6. The method of claim 1 , wherein the plant leaf is a potato plant leaf.

7. 10. The method of claim 1, further comprising applying a fertilizer treatment to a plant based at least in part on the nutritional value of the refined petiole.

8. A non-transitory computer-readable medium comprising instructions executable by a processor, the instructions, when executed, causing the processor to: storing leaf spectral data from a plant leaf in a memory; using the processor to convert the leaf spectral data into a preliminary petiole nutritional value by inputting the leaf spectral data into one or more linear machine learning models trained to provide an output of petiole nutritional value based on an input of the leaf spectral data; receiving an output from the one or more linear machine learning models comprising the preliminary petiole nutritional value; using the processor to refine the preliminary petiole nutritional value by inputting the preliminary petiole nutritional value into one or more non-linear machine learning models trained to provide an output of refined petiole nutritional value based on an input of the preliminary petiole nutritional value; receiving an output from the one or more nonlinear machine learning models comprising the refined nutritional value of the petiole; outputting a plant nutrient report indicating the nutritional value of the refined petiole; A non-transitory computer-readable medium configured to:

9. 10. The non-transitory computer-readable medium of claim 8, wherein the one or more non-linear machine learning models are trained to model interrelationships in the nutrient uptake profiles of different plant nutrients.

10. 10. The non-transitory computer-readable medium of claim 8, wherein the instructions, when executed, configure the processor to receive the spectral data taken from a leaf of a plant using a spectrophotometer.

11. 10. The non-transitory computer-readable medium of claim 8, wherein each of the one or more linear machine learning models is a Lasso regression model.

12. 10. The non-transitory computer-readable medium of claim 8, wherein each of the one or more non-linear machine learning models is a multivariate regression model.

13. 9. The non-transitory computer-readable medium of claim 8, wherein each of the one or more non-linear machine learning models is one of a support vector regression (SVR) model, a random forest (RF) model, and an extreme gradient boosting (XGB) model.

14. 10. The non-transitory computer readable medium of claim 8, wherein the instructions, when executed, configure the processor to apply a fertilizer treatment to a plant based at least in part on the refined nutrient value of the petiole.

15. 1. A system for determining the nutritional value of a petiole refined based on spectral data of the petiole from a plant leaf, comprising: a memory configured to store leaf spectral data from a plant leaf; converting the leaf spectral data into a preliminary petiole nutritional value by inputting the leaf spectral data into one or more linear machine learning models trained to provide an output of petiole nutritional value based on an input of the leaf spectral data; receiving an output from the one or more linear machine learning models comprising the preliminary petiole nutritional value; refining the preliminary petiole nutritional value by inputting the preliminary petiole nutritional value to one or more non-linear machine learning models trained to provide an output of refined petiole nutritional value based on the input of the preliminary petiole nutritional value; receiving an output from the one or more nonlinear machine learning models comprising the refined nutritional value of the petiole; outputting a plant nutrient report indicating the nutritional value of the refined petiole; a processor configured to A system comprising:

16. 16. The system of claim 15, wherein the one or more nonlinear machine learning models are trained to model interrelationships in the nutrient uptake profiles of different plant nutrients.

17. 16. The system of claim 15, wherein the processor is further configured to receive the spectral data taken from a plant leaf using a spectrophotometer.

18. 16. The system of claim 15, wherein each of the one or more nonlinear machine learning models is a multivariate regression model.

19. 16. The system of claim 15, wherein each of the one or more nonlinear machine learning models is one of a support vector regression (SVR) model, a random forest (RF) model, and an extreme gradient boosting (XGB) model.

20. 16. The system of claim 15, wherein the processor is further configured to apply a fertilizer treatment to a plant based at least in part on the refined nutrient value of the petiole.