GB2609614A - Variable rate herbicide application maps using UAV images - Google Patents

Abstract

A method and system for deploying weedkiller over an area of interest are disclosed. An image (for example an aerial/UAV image) is obtained. Image analysis is carried out, identifying the degree of invasive plant species coverage and gridding the image into cells 703. A threshold is set and cells are subdivided according to whether the weed percentage exceeds the threshold or not. A final map 700 is output, showing differentially the different types of cells. The cells may be split into different weed coverage bands, each band requiring different types of treatment. The invasiveness ranges in each bin may be identical. Different types or strengths of chemical may be applied at the same time. The total amount of weedkiller required to bring down the weeds below a uniform knockdown level may be calculated as well as the optimal cell set for weedkiller deployment when the available amount of chemical is limited. The images may be an orthomosaic. The different cells may adopt a colour scale proportional to the weed coverage, which can be displayed as a colour depth. The system may use a GPS receiver to determine when a cell boundary is crossed.

Description

Variable rate herbicide application maps using UAV images

Field of Invention

This invention relates to agricultural technology. More specifically, it relates to generating variable rate herbicide application maps using UAV images.

Background

Aerial images of the earth's surface may be taken using satellites orbiting in space. Such images usually comprise a plurality of frequency bands, including bands for the visible spectrum. ESAs Sentinel-2 satellites aim to image the whole of the earth's surface at least once per week in thirteen different bands (including bands for red, green, blue, IVIR) Satellites also image in different bands, for example ultraviolet, short-wave infrared (SW1R), etc, bands.

Aerial images (e.g. images taken by satellites) may be used to analyse vegetation cover, for example by forests and agricultural land. Analysis of crops in agricultural fields may be used to identify crop cover, maturity, and health. Collection of crop cover and/or maturity data over time allows the early identification of several different types of stress in the crop, provides valuable insights on the expected yield, and may enable prediction of when the crop will be ready to harvest. The analysis of crops may involve an index, for example the green area index (Berry PM, Spink J H. 2006. A physiological analysis of oilseed rape yields: Past and future. The Journal of Agricultural Science 144:381-392) or the normalised difference vegetation index (NDVI) (Rouse, SW, Haas, R.H., Scheel, TA., and Deering, D.W. (1974) Monitoring Vegetation Systems in the Great Plains with ERTS.' Proceedings, 3rd Earth Resource Technology Satellite (ERTS) Symposium, vol. 1, p. 48-62).

Typically, if a field is found to have vegetation cover, weedkiller will be applied to the entire field. This may lead to severe overspraying, since there will be parts of the field that have less vegetation cover than others. Indeed, many parts of the field may have no vegetation cover at all, which means that applying weedkiller to them is wasteful. This provides the challenge of ensuring that vegetation across a field is brought down to a desired level while ensuring an optimum use of resources.

Summary of invention

A method for deploying weedkiller over an area of interest is provided. The method comprises: obtaining an image of the area of interest; performing image analysis to identify degrees of weed coverage across the area of interest and applying a grid map made up of grid cells, thereby providing a gridded weed map of the area of interest; setting a threshold; calculating, from the threshold and the gridded weed map, those cells of weed growth exceeding the threshold; and generating a map of the area of interest showing, differentially, first cells of weed growth exceeding the threshold and second cells of weed growth not exceeding the threshold.

Preferably, the method further comprises splitting the cells into bands depending on their degree of weed coverage, wherein cells in different bands require different treatments. This helps to ensure that each cell receives an appropriate amount of weedkiller.

Preferably, the number of bands is any integer between 2 and 9. This also helps to ensure that each cell receives an appropriate amount of weedkiller.

Preferably, each band has an identical range of weed coverage. This also helps to ensure that each cell receives an appropriate amount of weedkiller.

Preferably, the different treatments are weedkillers of different types and/or strength deployed contemporaneously. This ensures that cells with varying degrees of weed coverage are catered for The threshold may comprise an upper threshold and/or a lower threshold. This enables the user to focus on the most important cells, rather than cells that evidently have too much or too little weed coverage.

Preferably, the method further comprises inputting a weedkiller type and/or strength. In the preferred embodiment, the type and strength are such as to give rise to the assumption that substantially all weed will be killed Preferably, the threshold relates to weed coverage at the time when the image was taken. This ensures that an accurate assessment of the weed coverage may be made.

Preferably, the method further comprises controlling a weedkiller deployment system across the area of interest, wherein the system is controlled by the map to deploy weedkiller differentially to the first and second cells. This ensures that each cell receives the correct amount of weedkiller.

Preferably, the system uses a GPS receiver to determine when a cell boundary is crossed and to switch between first and second states of deployment This ensures that the weedkiller is

deployed accurately across the field

Preferably, the method further comprises calculating a total amount of weedkiller required to bring the weeds below a uniform knockdown level. This ensures that a user does not use too much or too little weedkiller.

Preferably, the method further comprises calculating, given a set amount of available weedkiller, an optimum set of cells into which to deploy the weedkiller. This allows an efficient route for a sprayer to be calculated and ensures that other resources are not wasted by making the sprayer travel large distances to reach very small, isolated areas of weed growth.

Preferably, the at least one aerial image undergoes an orthomosaicking procedure. This produces a map that is geometrically more accurate, which helps to generate more reliable results Preferably, each cell represents an area of 5x5 metres. This is a useful size for a sprayer such as one attached to a tractor.

Preferably, the method further comprises colouring the map to show which cells have weed growth exceeding the threshold and which cells have weed growth not exceeding the threshold This enables a user to easily see which cells require attention Preferably, the depth of colour of the map varies with weed coverage. This provides a user with an even clearer view of which areas of the field may require more or less chemical.

A computer implemented plants analysis apparatus for deploying weedkiller over an area of interest is provided. The apparatus comprises: an input device for receiving at least one aerial image of the area of interest; means for performing image analysis to identify degrees of weed coverage across the area of interest; a mapping module for dividing the area of interest into multiple cells and calculating the weed coverage per cell; and an output device for displaying results in the form of a map of the area of interest with colour or scale for each cell corresponding to the weed coverage in the cell.

Preferably, the output device shows which areas of the field have weed coverage falling in different bands depending on their degree of weed coverage. This enables a user to clearly see which areas of the field require more or less chemical.

Preferably, the depth of colour of the map varies with weed coverage. This provides a user with an even clearer view of which areas of the field may require more or less chemical Preferably, the apparatus further comprises an automated weedkiller application device for receiving the results and selectively deploying weedkiller according to the indication. This ensures that each cell receives the correct amount of weedkiller.

Preferably, the apparatus has a location indicating device and deployment of weedkiller is dependent on the location within the map This ensures that the weedkiller is deployed accurately across the field.

Brief description of the drawings

Fig. 1 is a flow diagram depicting a method of analysing plants in a field.

Fig. 2 is a flow diagram depicting an orthomosaicking procedure.

Fig. 3 is a flow diagram depicting a post-processing and output display procedure Fig. 4 is a flow diagram depicting part of the post-processing and output display procedure of Fig. 3 Fig. 5 is a histogram showing an example of cells divided into bands.

Fig. 6 is an alternative histogram showing an alternative manner of dividing cells into bands.

Fig. 7 is a diagram showing a gridded weed map.

Fig. 8 is a flow diagram depicting a method of determining which cells to colour in a map of

afield.

Fig. 9 is a diagram showing a computer system

Detailed description

Referring to Fig. 1, a computer implemented program 100 for analysing plants is illustrated. UAV raw images 101 may be defined as at least one unprocessed aerial image of one or more fields obtained by an unmanned aerial vehicle (UAV) during a flight. UAV flight metadata 102 is geolocation and image quality metadata related to the flight of the UAV. Field boundaries 103 specify the geographic extent of each field imaged during the flight. Field boundaries are defined and stored oftline before the first UAV flight of the one or more fields. One or more fields may be imaged during a single flight.

Orthomosaicking module 104 is a software module used to generate an orthomosaic, which is a geometrically accurate aerial image that is composed of many individual still images that are stitched together and orthorectified (geometrically transformed to produce a top-down area view). UAV raw images 101, UAV flight metadata 102 and field boundaries 103 are all inputs for the orthomosaicking module 104, which is described in greater detail in Fig. 2.

Algorithm 105 is used to determine weed coverage across a field. The orthomosaic produced by orthomosaicking module 104 may be fed into the algorithm 105, which then analyses the orthomosaic image. This analysis may be chromatic analysis, wherein the detection of a certain colour or range or set of colours is linked with the presence of weeds or other plants.

For example, the analysis may be normalized difference vegetation index (NDVI) analysis, wherein the presence of green vegetation is determined on a pixel by pixel basis An NDVI value may be determined for each pixel. The output of the algorithm 105 may be a value or set of values per pixel.

Different parameters 106 may be fed into algorithm 105 depending on what type of camera has been used to obtain the UAV raw images 101.

Post-processing module 107 refers to a step of additional processing to an output of the algorithm 105. For example, the post-processing 107 may be performed on the output from the algorithm 105. For the post-processing, the data may be aggregated in cells of 5x5 metres or other suitable cell size. For each cell of 5x5 metres, the highest value that the algorithm 105 gave for any pixel within that cell is used as the value for that cell. This means that if any pixel within a cell has a weed or any other plant in it, the cell will be coloured. Colouring to different degrees is optional and is described below. The rest of the post-processing may then be performed on this updated image.

Display module 108 may display to a user the output of the post-processing module 107. The output may be in the form of a colour-coded map, as will be discussed in Figs 7 and 8.

In operation, UAV raw images 101 and UAV flight metadata 102 are all taken from a single flight. Field boundaries 103 are usually defined offline, rather than during the flight. The algorithm 105 is then applied to the image within the field boundary. Parameters 106 may be fed into the algorithm 105. The data from the algorithm 105 may be aggregated in areas of 5x5 metres to produce an updated image. Post-processing 107 may then be performed on the updated image. The post-processing 107 is described in greater detail in Fig. 3. The output of the post-processing module 107 may be displayed through a display module 108.

Referring now to Fig. 2, an orthomosaicking process 200 is described. The orthomosaicking process 200 may be performed by orthomosaicking module 104 of Fig. 1. Starting at block 201, the flight path of the UAV is reconstructed. This may be performed using the UAV flight metadata 102 from Fig. 1. At 202, any raw images that lie within a particular field or fields are found. The raw images may be the UAV raw images 101 and the field boundaries 103 may be used to determine whether the raw images lie within a particular field or fields.

At 203, the field coverage percentage is estimated. Specifically, for a given field, the percentage of the total field area that is covered by the raw images is estimated. If the field coverage percentage is less than 80 percent, the process is aborted. If the field coverage percentage is greater than or equal to 80 percent, the process continues to step 204.

At step 204, any overlapping pairs of images are determined. These pairs of images will be stitched together.

At step 205, the orthomosaicking software is run to produce an orthomosaic.

Referring now to Fig. 3, details of the post-processing module 107 and display module 108 of Fig. I are shown. Gridded weed map 301 is a map of a field split into cells of a specific size.

The gridded weed map 301 may be a map outputted by the algorithm 105 from Fig. Ito which a grid map has been applied, or may be a direct output from the algorithm 105 itself As shown in Fig. 3, the cells may represent an area of 5x5 metres, although the cells may represent areas of larger or smaller sizes, for example from 2x2 metres to 10x10 metres. The cells may be square cells, or may be other shapes, such as rectangular or hexagonal cells. The map 301 may show the extent of weed growth.

In Figs 1, 2 and 3, each box is a module of computer code performed by a processor, having memory that stores computer code and parameters In an alternative arrangement, by using suitable Mask R-CNN software, the map 301 may show the number of individual weeds in each cell and the sizes of the weeds in each cell.

Weed coverage threshold 302 is a threshold set by the user related to how much of a cell must be covered by weeds in order for it to fall into a particular category and subsequently how much chemical should be applied to it. The weed coverage threshold 302 may be inputted as a fraction, a decimal or a percentage The weed coverage threshold 302 may be inputted through an interactive slider that ranges from 0 to 1 (or 0% to 100%) The weed coverage threshold 302 may comprise a lower threshold and/or an upper threshold.

The default lower threshold may be 0 (0% if working in percentages) and the default upper threshold may be 1(100% if working in percentages) A user may keep the upper and lower thresholds at their default levels, or they may alter one or both of the upper and lower thresholds using the slider.

Number of bands 303 is a parameter inputted by the user that determines how many bands the cells should be split into. Each band represents a particular range of weed coverage and, correspondingly, a particular amount of chemical that should be applied. The range of each band is calculated based on the weed coverage threshold 302. To determine the range of each band, the difference between the upper and lower thresholds is preferably calculated and divided by the number of bands 303. In this way, each band has an identical range of weed coverage, but with different maximum and minimum values. For example, two bands may be selected. The weed coverage threshold 302 may have been set to the default option, with the lower threshold at 0 and the upper threshold at 1. The difference between the two thresholds is therefore 1 and the range of each band is therefore 0.5. One band will therefore be for cells with weed coverage between 0 and 0.5 (0%-50%), whereas the other band will be for cells with weed coverage between 0.5 and 1 (50%-100%). The number of bands is preferably any integer between 2 and 9 (inclusive), but more bands are possible. In another example, three bands may be selected. The weed coverage threshold 302 may have been set with a lower threshold of 0.1 and an upper threshold of 0.7. The range of each band is therefore 0.2. One band will therefore be for cells with weed coverage between 0.1 and 0.3, another band will be for cells with weed coverage between 0.3 and 0.5, and the final band will be for cells with weed coverage between 0.5 and 0.7.

Typically, the lowest band will have the lowest amount of chemical applied (often none at all). Higher bands will have greater amounts of chemical applied.

Preferably, any cells with coverage below the lower threshold will not be sprayed.

Preferably, any cells with coverage above the upper threshold will be sprayed with the same amount of chemical as applied to cells in the highest band, because experience indicates that they have similar sized roots and will be knocked back to the same degree as the weeds in the highest band These measures allow for reduced spraying at both the lower and upper levels.

Alternatively, the weed coverage threshold 302 may relate to a desired covering of weeds after a specified period of time. In this case, the weed coverage threshold 302 only requires a lower threshold to be set, regardless of the number of bands 303. For example, a user may wish for all cells to have a weed coverage of 10% or less in three weeks time. For each cell, the appropriate treatment (i.e. amount of chemical needed to bring the weed coverage to below 10% in three weeks time) may then be determined and each cell may then be placed into the appropriate band. For example, a user may set the lower threshold of the weed coverage threshold 302 as being 10% in three weeks time and the number of bands 303 as three. For cells that, at the time of input, have a large weed covering, a first treatment may be required, corresponding to a first of the three bands. This first treatment may represent a large quantity of chemical being applied (for example, one litre per hectare). For cells that, at the time of input, have a smaller weed covering (but one that will still be above 10% in three weeks time), a second treatment may be required, corresponding to a second of the three bands. This second treatment may represent a smaller quantity of chemical being applied (for example, half a litre per hectare). For cells that have no weed covering (or such a small covering that in three weeks time it will still be below 10%), no treatment is required (corresponding to a third of the three bands). Each treatment may refer to application of a certain type and/or strength of weedkiller deployed contemporaneously. It is to be understood that the values provided above are simply examples and that many other values/ranges are possible.

Pre-output visualisation 304 is a rendering of what will eventually be outputted by an output layer 307. This visualisation is formed using gridded weed map 301, weed coverage threshold 302 and number of bands 302as inputs, and may be used as an intermediate visual output for a user. In this way, a user may see what their output will look like and may amend the inputs if they are not satisfied with this output. The pre-output visualisation may appear as a gridded map with certain cells coloured in.

The generate application module 305 is used to create an output map based on the gridded weed map 301, weed coverage threshold 302, number of bands 303 and application volume and product rates 306. This output map may be outputted through the output layer 307 Application volume and product rates 306 represent quantities of a chemical that should be applied to each cell of a particular band, depending on the size/number of weeds in that cell. These quantities may be set by a user and may have units of litres per hectare (1/ha). The chemical may be a weed-killing chemical. Bands for cells with larger weed coverage will have larger application volume and/or concentration and/or product rates 306 (i.e. more chemical applied) whereas bands for cells with smaller weed coverage will have smaller application volume and/or concentration and/or product rates 306 (i.e. less chemical applied). Preferably, all applications in all bands have the same or similar viscosity. Application volume and product rates 306 may represent a particular weedkiller type and/or strength and may be inputted by the user.

In operation, a gridded weed map 301 is inputted. A user may then set a weed coverage threshold 302 and number of bands 303. In real time, a pre-output visualisation 304 is provided, which automatically updates whenever either of the weed coverage threshold 302 or number of bands 303 are changed. This visualisation shows a map of the field covered by the gridded weed map 301. The cells of the map may be coloured based on the user's inputs. When the user inputs a particular weed coverage threshold, any cells that fall into particular bands as a result of them exceeding or falling short of a threshold will automatically be shown as being coloured in by the pre-output visualisation 304. Different bands may be represented by different colours or by different shades of the same colour. The band that includes cells below the lowest threshold may not be coloured at all.

It is to be understood that since the pre-output visualisation 304 is a real-time rendering, any of the inputs may be changed at any time.

Additionally, a user may alter the baseline contrast of the pre-output visualisation 304. In this way, the weeds may appear more clearly on the map itself This is purely for making the weeds easier for the user to see and has no effect on the colouring of the map in accordance with the bands and thresholds.

A user may then generate the application through the generate application module 305. This produces an output map through the output layer 307. When producing this map, application volume, concentration and/or product rates 306 are inputted for each band. The operation of the generate application module 306 is described in greater detail in Fig. 4. The output map may appear identical to the pre-output visualisation 305. The output map may be in the form of a coloured, gridded map, which is described in greater detail in Fig. 7.

The output map may show a user which cells of a field are in which band and which cells require a chemical to be applied. It may also show a user how many hectares of the field fall into each band and how much chemical will be needed depending on the inputted parameters. The output map may therefore show the total amount of chemical required to bring the weeds below a uniform "knockdown" level (i.e. a suitably low level of weed coverage). At this stage, the application volume and product rates 303 may be altered. For example, a user may have set two bands, one of which requires one litre of chemical to be applied per hectare.

When presented with the total number of hectares falling into that band and the total amount of chemical required, the user may realise that they do not have enough of the chemical to achieve this and may lower the rate. For example, the rate may be lowered to 0.8 litres per hectare.

The generation of the output map may also be determined based on a set amount of chemical (weedkiller) available. A user may input the total amount of weedkiller that can be applied to the area of interest. A level of deployment of chemical required to achieve a uniform level of weed coverage across the field can then be calculated and displayed in the output map.

Still referring to Fig. 3, it is to be understood that the generation of the output map may also be based on the most efficient deployment of chemical based on distances between cells, as well as on weed coverage of the cells and amount of weedkiller available For example, if the output map shows that the vast majority of cells in a band requiring a large amount of chemical are located on one side of the field and that this band also comprises one outlying cell at the opposite side of the field, it may be determined that travelling all the way across the field to apply the chemical to this one cell may not be a particularly efficient use of resources. The output map may therefore re-categorise this cell into another band, or omit it entirely, if it is determined that this is the case. In this way, given a set amount of available weedkiller, an optimum set of cells into which to deploy the weedkiller can be determined.

The output map may be downloaded by the user and used to determine which areas of a field should receive a particular quantity of chemical. A weedkiller deployment system may be used to apply the chemical in accordance with the output map. The system may comprise an automated weedkiller application device for receiving the map and selectively deploying weedkiller according to the map. In such a system, the map may be GPS coordinated with an external spraying device to ensure that the subsequent spraying of the chemical directly aligns with the areas marked on the map. For example, the map may directly control a valve of sprayer drawn by a tractor (or carried by an autonomous vehicle or drone, or attached to an irrigation system), where the valve opens and closes (or changes its flow rate) dependent on the output of a GPS receiver (or other location indicating device) as the GPS receiver passes from cell to cell according to the map.

For example, if the output map divides the cells into first cells requiring a first treatment and second cells requiring a second treatment, the weedkiller deployment system may be controlled by the map to deploy weedkiller differentially to the first and second cells. The system may use a GPS receiver to determine when a cell boundary is crossed and to switch between first and second states of deployment as necessary. Here, the first state of deployment corresponds to the first treatment and the second state of deployment corresponds to the second treatment. As has been discussed, it is to be understood that one or more of the treatments may be to apply no chemical at all.

Referring now to Fig. 4, the generate application module 306 from Fig. 3 is described in more detail. Starting at step 401, the module receives data from the user. This may be in the form of the gridded weed map 301, the weed coverage threshold 302, the number of bands 303 and the application volume and product rates 306.

At step 402, this data is retrieved to use as basis for the application.

At step 403, the data is simplified based on user input to create zones/bands/levels. It is to be understood that the terms "zones", "bands" and "levels" are interchangeable For example, if the user has requested two bands to be set, the data is simplified such that it can be split into these two bands.

At step 404, application volume and product rates may be generated based on user input for each of the simplified zones/bands/levels. For example, if the user has selected one band to receive one litre of chemical per hectare and another band to receive zero litres of chemical per hectare, the module will apply these rates At step 405, the application is saved and sent to the frontend (i.e. he output layer 307 of Fig 10 3) Fig. 5 shows an example histogram in which the data has been split into bands of equal width (i.e. range). As has been discussed, the range of each band may be calculated using the following equation: Range - Upper threshold -Lower threshold Number of bands Referring now to Fig. 5, an example histogram 500 is shown to aid in understanding how the data is divided into bands. The histogram 500 has weed coverage on its x-axis 501, shown as ranging from 0 (no coverage) through to I (complete coverage). The histogram 500 has number of cells on its y-axis 502. No numbers are shown on the y-axis 502 as this is dependent on the total number of cells across the image, which depends on the size of the image and on the size of the cells.

The plotted data 503 represents all the cells in the image plotted based on the weed coverage of each cell. The taller a column within the plotted data 503, the more cells within the coverage range represented by that column there are. Here, the width of each column (i.e. the coverage range) is 0.05, although this is just an example and it should be understood that larger or smaller widths are possible. Here, the example histogram 500 has a normal distribution, although it is to be understood that the distribution depends on the weed

coverage across the field.

As mentioned, a user may set a weed coverage threshold 302, which may comprise a lower and an upper threshold. In Fig. 5, the lower threshold 504 has been set as 0.1 and the upper threshold 505 has been set as 0.7. The number of bands 303 has been set as 3. As has been discussed, this means that the lower band 506 is for cells with weed coverage of 0.1-0.3, the middle band is for cells with weed coverage of 0.3-0.5 and the upper band is for cells with weed coverage of 0.5-0.7. Cells with weed coverage greater than the upper threshold may not be of particular concern to the user and will typically receive the same amount of chemical as cells in the upper band 508. Cells with weed coverage lower than the lower threshold may also not be of particular concern to the user and will typically receive no chemical.

Referring now to Fig. 6, an example histogram 600 is shown an alternative way of dividing the data into bands. The histogram 600 has weed coverage on its x-axis 601, shown as ranging from 0 (no coverage) through to I (complete coverage). The histogram 600 has number of cells on its y-axis 602. No numbers are shown on the y-axis 602 as this is dependent on the total number of cells across the image, which depends on the size of the image and on the size of the cells.

The plotted data 603 represents all the cells in the image plotted based on the weed coverage of each cell. The taller a column within the plotted data 603, the more cells within the coverage range represented by that column there are. Here, the width of each column (i.e. the coverage range) is 0.05, although this is just an example and it should be understood that larger or smaller widths are possible. Here, the example histogram 600 has a normal distribution, although it is to be understood that the distribution depends on the weed coverage across the field.

As mentioned, a user may set a weed coverage threshold 302, which may comprise a lower and an upper threshold. In Fig. 6, the lower threshold 604 has been set as 0.1 and the upper threshold 605 has been set as 0.9. The number of bands 303 has been set as 5. In this alternative way of dividing the data into bands, each of the bands 606, 607, 608, 609 and 610 has approximately the same number area, representing approximately the same number of cells within the band. This is in contrast to Fig. 5, in which each band has an equal range, but may not have the same area (and subsequently may not have the same number of cells in each band). The bands 606, 607, 608, 609 and 610 may therefore have different ranges. It is to be understood that the areas of each band need not be totally equal -it is sufficient for them to be approximately equal. Cells with weed coverage greater than the upper threshold may not be of particular concern to the user and will typically receive the same amount of chemical as cells in the upper band 610. Cells with weed coverage lower than the lower threshold may also not be of particular concern to the user and will typically receive no chemical.

An example of an output map showing which cells fall into which bands is shown in Fig. 7.

In this way, a user is presented with a clear and simple representation of which areas of the field have weeds growing in them. These areas may be split into categories such as: (a) large 10 weed coverage; (b) medium weed coverage; (c) small weed coverage; and (d) no weed coverage. More or fewer categories are possible.

Depending on the number of bands, the depth of colour may vary with weed coverage. For example, a larger weed coverage in a cell may be represented by a deeper colour and a smaller weed coverage by a lighter colour. In this context, depth of colour is intended to include depth of greyscale.

As another example, category (a) can be deep orange, category (b) medium orange and category (c) light orange. Category (d) can be very light orange or a different colour such as brown or red. Category (d) may be represented by no colour at all.

If only two bands are set, then depth of colour may not be used. In this case, one band may be represented by one colour (for example, orange) and the other band may be represented by another colour (or no colour).

From such a map, a user can readily deploy sprayers to cells in particular categories, depending on the number of bands and the thresholds set. For example, sprayers may be deployed to cells in categories (a) and (b), but not to cells in categories (c) and (d). In this way, chemicals are not wasted on cells in categories (c) and (d), thus saving on cost and reducing phosphate and nitrate pollution. Reference to "deploying' a sprayer may include activating a localized region of a pre-deployed spraying system.

Referring still to Fig. 7, an example of an output map 700 is shown. The output map 700 may be the output map from the output layer 307 of Fig. 3. Output map 700 comprises an image of an area 701, which may cover part of a field, a field, or multiple fields. A grid 702 may be applied to the image of the area 701, which splits the image 701 into cells 703, as has already been discussed with regard to gridded weed map 301 of Fig. 3. As already mentioned, the dimensions of each cell 703 and hence the area covered by each cell 703 may be set by the user. The cells 703 in Fig 7 are shown as being square, but may be hexagonal, rectangular, or another shape.

The image of an area may 701 may also show areas of weed coverage 704. The areas of weed coverage 704 may be outputted by the algorithm 106 of Fig. 1 and may show the user where in the image 701 weeds are located. From this, the weed coverage of each cell may be determined. As has been discussed, the weed coverage may be determined through chromatic analysis.

As has been discussed, each cell can be coloured depending on which band it falls into. In Fig 7, three bands have been set. Cells with weed coverage in the highest band are shown as being shaded with a dark grey, cells with weed coverage in the middle band are shown as being shaded with a lighter grey, whereas cells with weed coverage in the lowest band are shown as being non-shaded.

Referring now to Fig. 8, a process 800 that may take place in both the pre-output visualisation 305 and the generate application module 306 is shown. The process 800 may be used to generate an output map showing where a particular chemical should be applied. A user may only wish to apply chemicals to areas that fall into particular bands. In Fig. 8, a situation in which only two bands have been set is shown, but it is to be understood that the same process could be applied to any number of bands.

At step 801, it is determined, for a particular cell, whether the weed coverage in that cell is above a threshold. It is to be understood that the threshold preferably relates to a present threshold (i.e. at that present moment) but may relate to a future threshold (e.g. in three weeks' time, given the present rate of growth). The threshold may be the weed coverage threshold 302 from Fig. 3. If the answer is yes, the cell is allocated to the higher band and the process moves to step 802, where an Apply_Chemical parameter is set to I. If the answer is no, the cell is allocated to the lower band and the process moves to step 803, where the Apply Chemical parameter is set to 0. This is performed for every cell. At step 804, a colour-coded grid is outputted. This colour-coded grid displays the cells for which the Apply_Chemical parameter is 1 as one colour and the areas for which the Apply_Chemical parameter is 0 as a different colour (or no colour). The resulting grid therefore shows the user which areas of the field require a chemical to be applied. This grid may be the output map from the output layer 307.

As mentioned earlier, the threshold may be set by a user. For example, a user may determine that cells with weed coverage above a threshold may need a chemical to be applied to them. These cells may then be coloured on the map.

In this way, a user can set their desired thresholds and, from the output, may be able to determine where in the field a certain chemical should be applied As an alternative to calculating present thresholds, calculations of future weed growth after 15 weedkiller deployment can be made (dependent on weedkiller type and strength and passage of time), and future thresholds can be set based on such projected growth (or lack thereof).

Referring now to Fig. 9, an example computer system 900 is shown in which the present invention, or portions thereof can be implemented as computer-readable code to program processing components of the computer system 900. Various embodiments of the invention are described in terms of this example computer system 900. For example, computer implemented program 100 of Fig. 1 can each be implemented in such a system 900. The methods illustrated by the flowcharts of Figs 2, 3, 4 and 8 can also be implemented in such a system 900. After reading this description, it will become apparent to a person skilled in the relevant art how to implement the invention using other computer systems and/or computer architectures. At least one input to the computer system 900 must be an aerial image.

Computer system 900 includes one or more processors, such as processor 902 Processor 902 can be a special purpose or a general-purpose processor. Processor 902 is connected to a communication infrastructure 901 (for example, a bus, or network). Computer system 900 also includes a user input interface 903 connected to one or more input devices 904 and a display interface 905 connected to one or more displays 906, which may be integrated input and display components. Input devices 904 may include, for example, a pointing device such as a mouse or touchpad, a keyboard, a touchscreen such as a resistive or capacitive touchscreen, etc. A computer display 907 (not shown in Fig. 9), in conjunction with display interface 905, can be used as display 110 shown in Fig. 1 and can display the results 700 shown in Fig. 7. Alternatively, the results 700 can be printed on paper using printer 909 through printer interface 908.

Computer system 900 also includes a main memory 910, preferably random access memory (RAM), and may also include a secondary memory 911. Secondary memory 911 may include, for example, a hard disk drive 912 (not shown in Fig. 9), a removable storage drive 913 (not shown in Fig. 9), flash memory, a memory stick, and/or any similar non-volatile storage mechanism or cloud memory. Either or both of main memory 910 and secondary memory 911 may include means for allowing computer programs or other instructions to be loaded into computer system 900. Examples of such means may include a program cartridge and cartridge interface (such as that found in video game devices), a removable memory chip (such as an EPROM, or PROM) or the like.

Computer system 900 may also include a communications interface 914. Communications interface 914 allows software and data to be transferred between computer system 900 and external devices 915. Communications interface 914 may include a modem, a network interface (such as an Ethernet card), a communications port, a PCMCIA slot and card, or the like. Communications interface 914 may be a wireless communications interface. External devices 915 may include a sprayer or sprayers, such as those attached to a tractor or tractors, drones or an irrigation system.

Various aspects of the present invention can be implemented by software and/or firmware (also called computer programs, instructions or computer control logic) to program programmable hardware, or hardware including special-purpose hardwired circuits such as application-specific integrated circuits (ASICs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), etc of the computer system 900, or a combination thereof Computer programs for use in implementing the techniques introduced here may be stored on a machine-readable storage medium and may be executed by one or more general-purpose or special-purpose programmable microprocessors.

Computer programs, model parameters and model training data for a trained model are stored in main memory 910 and/or secondary memory 911. It will also be appreciated that the model stored in these memories can be trained (and fixed) or adaptive (and susceptible to further training). Computer programs may also be received via communications interface 914. Such computer programs, when executed, enable computer system 900 to implement the present invention as described herein. In particular, the computer programs, when executed, enable processor 902 to implement the processes of the present invention, such as the steps in the methods illustrated by the flowcharts of Figs 2, 3, 4 and 7, and the system component architectures of Fig. 1 described above. Accordingly, such computer programs represent controllers of the computer system 900. Where the invention is implemented using software, the software may be stored in a computer program product and loaded into computer system 900 using removable storage drive 913, interface 903, hard drive 912, or communications interface 914.

Embodiments of the invention employ any computer useable or readable medium, known now or in the future. Examples of computer useable mediums include, but are not limited to, primary storage devices (e.g., any type of random access memory), secondary storage devices (e.g., hard drives, floppy disks, CD ROMS, ZIP disks, tapes, magnetic storage devices, optical storage devices, MEMS, nano-technological storage device, etc.), and communication mediums (e.g., wired and wireless communications networks, local area networks, wide area networks, intranets, etc.).

It will be understood that embodiments of the present invention are described herein by way of example only, and that various changes and modifications may be made without departing from the scope of the invention.

References in this specification to "one embodiment" are not necessarily all referring to the same embodiment, nor are separate or alternative embodiments mutually exclusive of other embodiments. In particular, it will be appreciated that aspects of the above described embodiments can be combined to form further embodiments. Similarly, various features are described which may be exhibited by some embodiments and not by others. Yet further alternative embodiments may be envisaged, which nevertheless fall within the scope of the following claims.

Claims (2)

1. Claims I. A method for deploying weedkiller over an area of interest comprising: obtaining an image of the area of interest; performing image analysis to identify degrees of weed coverage across the area of interest and applying a grid map made up of grid cells, thereby providing a gridded weed map of the area of interest; setting a threshold; calculating, from the threshold and the gridded weed map, those cells of weed growth exceeding the threshold; and generating a map of the area of interest showing, differentially, first cells of weed growth exceeding the threshold and second cells of weed growth not exceeding the threshold.
2. 2. The method of claim 1, further comprising splitting the cells into bands depending on their degree of weed coverage, wherein cells in different bands require different treatments The method of claim 2, wherein the number of bands is any integer between 2 and 9.4. The method of claim 2 or 3, wherein each band has an identical range of weed coverage.5. The method of any one of claims 2-4, wherein the different treatments are weedkillers of different types and/or strength deployed contemporaneously.6. The method of any one of the preceding claims, wherein the threshold comprises an upper threshold and/or a lower threshold.7 The method of any one of the preceding claims, further comprising inputting a weedkiller type and/or strength.8. The method of any one of the preceding claims, further comprising controlling a weedkiller deployment system across the area of interest, wherein the system is controlled by the map to deploy weedkiller differentially to the first and second cells 9. The method of claim 8, wherein the system uses a UPS receiver to determine when a cell boundary is crossed and to switch between first and second states of deployment.10. The method of any one of the preceding claims, further comprising calculating a total amount of weedkiller required to bring the weeds below a uniform knockdown level.11. The method of any one of the preceding claims, further comprising, calculating, given a set amount of available weedkiller, an optimum set of cells into which to deploy the weedkiller.12. The method of any one of the preceding claims, wherein the at least one aerial image undergoes an orthomosaicking procedure.13. The method of any one of the preceding claims, wherein each cell represents an area of 5x5 metres.14. The method of any one of the preceding claims, further comprising colouring the map to show which cells have weed growth exceeding the threshold and which cells have weed growth not exceeding the threshold.The method of claim 14, wherein the depth of colour of the map varies with weed coverage.16. A computer implemented plants analysis apparatus for deploying weedkiller over an area of interest, comprising: an input device for receiving at least one aerial image of the area of interest; means for performing image analysis to identify degrees of weed coverage across the area of interest; a mapping module for dividing the area of interest into multiple cells and calculating the weed coverage per cell; and an output device for displaying results in the form of a map of the area of interest with an indication for each cell corresponding to the weed coverage in the cell.17 The apparatus of claim 16, wherein the output device shows which areas of the field have weed coverage falling in different bands depending on their degree of weed coverage 18. The apparatus of claim 16 or 17, wherein the indication is a depth of colour that varies with weed coverage.19. The apparatus of any of claims 16-18, further comprising an automated weedkiller application device for receiving the results and selectively deploying weedkiller according to the indication.20. The apparatus of claim 19, wherein the automated weedkiller application device has a location indicating device and deployment of weedkiller is dependent on the location within the map.