KR102570846B1 - Field compost detection method and system capable of pollutant management - Google Patents

Abstract

The present invention comprises the steps of obtaining an orthoimage associated with a target area;
inputting the orthoimage to a pre-learned deep learning model, and acquiring a region of interest, which is an area where compost in the field exists, in the orthoimage;
calculating current status information related to the compost in the open field to determine a priority for managing the compost in the open field based on at least one of the orthoimage and the region of interest; and
It provides a method for detecting open compost, including the step of outputting the calculated status information.

Description

Field compost detection method and system capable of managing pollution sources {FIELD COMPOST DETECTION METHOD AND SYSTEM CAPABLE OF POLLUTANT MANAGEMENT}

The present invention relates to a method and system for detecting open compost capable of managing pollution sources.

While compost has value as a source of nutrients for crops in farmland, it can also act as a pollutant that adversely affects the aquatic environment as farm wastewater if not properly managed. Accordingly, compost stored in farmland is pointed out as a non-point source of pollution, but there is a limit to grasping the amount and distribution of compost in open storage. difficulties follow.

Therefore, since these non-point pollutants can directly affect water quality during rainfall, proper management is required by identifying the distribution and loading of yard compost. To this end, conventionally, a lot of manpower and time are required as a researcher directly records information related to open field compost in the field or uses spatial data such as a land cover map to identify open field compost.

In recent years, monitoring methods using drone images, aerial images, and satellite images have been used as efficient survey methods for large areas such as water systems. it is possible to acquire

In this regard, in addition to improving computer performance, high recognition rate and accuracy have been shown in the field of object recognition using deep learning technology, which is a sub-field of image processing and computer vision.

Therefore, in order to effectively manage and improve the water environment of the water system, among the 4th industrial technologies, open-air compost detection technology using drones and artificial intelligence is applied to accurately and quickly identify the current status of pollutants and manage continuous monitoring data. system is required.

On the other hand, the present invention is a part of the Nakdong River Water System Environmental Basic Survey Project (2020.07. ~ 2023.03.) by the Ministry of Environment and the Nakdong River Water System Management Committee, and a water system pollution source detection and water environment management plan using drones and artificial intelligence (Nakdong River Water System No. 2020-05 ) is about.

The present invention relates to a method and system for detecting compost in an open field from an image of a target area.

In addition, the present invention relates to a method and system for detecting field compost that provides information related to field compost so that a user can efficiently manage the field compost by determining the state of the detected field compost.

In order to solve the above problems, a method for detecting compost in an open field according to the present invention includes the steps of acquiring an orthoimage associated with a target area; inputting the orthoimage to a pre-learned deep learning model, and acquiring a region of interest, which is an area where compost in the field exists, in the orthoimage; calculating current status information related to the compost in the open field to determine a priority for managing the compost in the open field based on at least one of the orthoimage and the region of interest; and outputting the calculated status information.

In addition, the open-field compost detection system according to the present invention includes a storage unit for storing an orthographic image related to a target area; and inputting the orthoimage to a pre-learned deep learning model to acquire a region of interest, which is an area in which compost in the field exists, in the orthoimage, and based on at least one of the orthoimage and the region of interest, the compost in the field and a control unit that calculates status information related to the open compost and outputs the calculated status information so as to determine a priority for managing.

In addition, a program stored in a computer-readable recording medium according to the present invention is executed by one or more processes in an electronic device and is stored in a computer-readable recording medium, and the program is obtaining a related orthoimage; inputting the orthoimage to a pre-learned deep learning model, and acquiring a region of interest, which is an area where compost in the field exists, in the orthoimage; calculating current status information related to the compost in the open field to determine a priority for managing the compost in the open field based on at least one of the orthoimage and the region of interest; and outputting the calculated status information; may include instructions for performing.

According to various embodiments of the present invention, a method and system for detecting open field compost may detect an area in which field compost exists in an orthographic image, and furthermore, by calculating information related to the detected field compost, the management state and Possibility of contamination by field manure can be predicted.

In addition, according to various embodiments of the present invention, the method and system for detecting compost in an open field output the current state of the compost in an open field detected from an orthographic image in various forms, and furthermore, generate and output monitoring information on the compost in an open field, so that the user can It can be configured to easily manage a plurality of yard composts present in various locations.

In addition, the field compost detection method and system according to the present invention determines the inspection status of the field compost and the management level of the management state of the field compost, thereby providing the user with information on the field compost that requires priority management. .

1 shows a yard compost detection system according to the present invention.
2 shows a deep learning model learning system arranged to train a deep learning model.
3 is a flowchart of a method for detecting compost in an open field according to the present invention.
4 is a flowchart of a deep learning model learning method for learning a deep learning model prepared in a field compost detection system.
5 illustrates an embodiment of a region of interest obtained using a deep learning model prepared in a field compost detection system.
6 shows a plurality of modules provided to calculate current status information in the yard compost detection system.
7-16 illustrate one embodiment using a yard compost detection system.

Hereinafter, the embodiments disclosed in this specification will be described in detail with reference to the accompanying drawings, but the same reference numerals will be assigned to the same or similar components regardless of reference numerals, and overlapping descriptions thereof will be omitted. The suffixes "module" and "unit" for components used in the following description are given or used together in consideration of ease of writing the specification, and do not have meanings or roles that are distinct from each other by themselves. In addition, in describing the embodiments disclosed in this specification, if it is determined that a detailed description of a related known technology may obscure the gist of the embodiment disclosed in this specification, the detailed description thereof will be omitted. In addition, the accompanying drawings are only for easy understanding of the embodiments disclosed in this specification, the technical idea disclosed in this specification is not limited by the accompanying drawings, and all changes included in the spirit and technical scope of the present invention , it should be understood to include equivalents or substitutes.

Terms including ordinal numbers, such as first and second, may be used to describe various components, but the components are not limited by the terms. These terms are only used for the purpose of distinguishing one component from another.

It is understood that when an element is referred to as being "connected" or "connected" to another element, it may be directly connected or connected to the other element, but other elements may exist in the middle. It should be. On the other hand, when an element is referred to as “directly connected” or “directly connected” to another element, it should be understood that no other element exists in the middle.

Singular expressions include plural expressions unless the context clearly dictates otherwise.

In this application, terms such as "comprise" or "have" are intended to designate that there is a feature, number, step, operation, component, part, or combination thereof described in the specification, but one or more other features It should be understood that the presence or addition of numbers, steps, operations, components, parts, or combinations thereof is not precluded.

1 shows a yard compost detection system according to the present invention. 2 shows a deep learning model learning system arranged to train a deep learning model.

Referring to FIG. 1 , the open field compost detection system 100 according to the present invention detects an area of interest in which the open field compost exists in an orthoimage 1 related to a target area, and detects the current state of the open field compost in the detected area of interest. Information (2) can be calculated and provided to the user.

Here, the target area may be an area in which current conditions such as distribution and loading of open-air compost are to be grasped. For example, the target area may include agricultural land. At this time, the target area may include a water system (for example, a river, a river, etc.) existing near farmland, and in this case, the open-field compost detection system 100 identifies the possibility of contamination of the water environment of the water system by the open-field compost. can be used to do

Compost is grass, straw, livestock manure, and the like, and may refer to manure used as a source of nutrition for crops in farmland. Therefore, open-air compost may refer to compost piled up in the vicinity of a target area such as farmland. At this time, open-air compost is recommended to be stored in a designated location, and for example, open-air compost may be recommended to be stored in private land. That is, open-air compost can be treated as an illegal pile-up when it is stored on state land managed by the state for various purposes of use.

Such open-air compost may be lost due to rain or the like, and a cover such as vinyl may be covered to prevent this. At this time, the material and color of the cover may be pre-specified, and through this, the field compost detection system 100 may determine that the area in which the field compost covered with vinyl exists in the orthoimage 1 related to the target area is also a region of interest. can be detected as

In this regard, the current status information 2 may include a plurality of different pieces of information required in the process of grasping and managing the status of compost in the open field. For example, the current state information 2 may include various pieces of information such as the area, loading amount, location, distance from a water system, state-owned land, and presence or absence of cover.

Meanwhile, the orthoimage 1 may be produced by applying a technique such as orthogonal deviation correction to drone images, aerial images, and satellite images. For example, the orthoimage 1 acquires a plurality of images taken in consideration of a previously designated altitude, route, image overlap rate, etc. through a drone, and obtains three-dimensional coordinates (eg, For example, latitude, longitude, altitude) into ground coordinates, feature points are extracted from each image, feature points representing the same point are matched through comparison between extracted feature points, and based on the matched feature points, a 2D image A set of is converted into a 3-dimensional point cloud, and can be generated based on 3-dimensional coordinate values determined for each point of the point cloud.

In this regard, the open field compost detection system 100 may further use a Digital Surface Model (DSM) along with the orthoimage 1 to generate current status information 2 related to the open field compost.

In this case, the numerical surface model may include information related to the altitude of the surface of the ground. Such a numerical surface model may be generated by generating a point cloud using drone images, aerial images, satellite images, and the like, and using the generated point clouds. Through this, the numerical surface model may include information related to height at each pixel.

On the other hand, each of the orthoimage and the numerical surface model is not limited to the above examples, and those manufactured using various conventionally known techniques may be used.

Meanwhile, the open-air compost detection system 100 may include a storage unit 110 , an input unit 130 , a control unit 150 and an output unit 170 .

The storage unit 110 may store information and commands necessary for the operation of the field compost detection system 100 . For example, the storage unit 110 may store the orthoimage 1 and the numerical surface model related to the target area.

In addition, the storage unit 110 may store a deep learning model previously trained to detect a region of interest in the orthoimage 1 .

In this case, the deep learning model may be learned so that, when the orthoimage 1 is input, a region of interest, which is a region in which open field compost exists, is output in the orthoimage 1 . For example, the deep learning model may be one learned by the deep learning model learning system 200 .

Referring to FIG. 2 , the deep learning model learning system 200 may include a storage unit 210, an input unit 230, a learning unit 250, and an output unit 270.

The storage unit 210 may store information and instructions necessary for the operation of the deep learning model learning system 200 . For example, the storage unit 210 may store a plurality of ortho images 11 for learning and a plurality of regions of interest 12 with correct answers.

Here, the orthoimage 11 for learning may be an orthoimage prepared to train the deep learning model, and the region of interest 12 of the correct answer is label data for the orthoimage 11 for learning, and in the orthoimage 11 for learning, It may be a representation of a region of interest, which is a region in which compost is present in the open field, in a preset form (eg, a black and white image).

For example, the correct answer region of interest 12 may be an annotation for the orthoimage 11 for learning. In this case, the annotation may be information added to the source code in a programming language.

A plurality of orthoimages 11 for learning and a plurality of regions of interest 12 with correct answers prepared to learn a deep learning model may be input by a user to the input unit 230 .

The learning unit 250 may train a deep learning model using a plurality of orthoimages 11 for learning and a plurality of regions of interest 12 for correct answers. For example, deep learning models can be trained using various techniques such as deep neural networks (DNNs), convolutional neural networks (CNNs), and recurrent neural networks (RNNs). can

In addition, the learning unit 250 may perform verification of the learned deep learning model using a plurality of pre-prepared orthoimages for testing and a plurality of regions of interest for testing. In this case, the learning unit 250 inputs a test orthoimage to the learned deep learning model, and when a region of interest corresponding to the test orthoimage is output, the learning unit 250 compares the output region of interest with the tested region of interest, thereby learning deep learning. The accuracy of the model can be calculated.

The output unit 270 may output a process in which the deep learning model is learned by the learning unit 250 . For example, the output unit 270 may output a region of interest obtained by inputting an orthoimage for testing to the trained deep learning model, and output a region of interest for testing corresponding to the orthoimage for testing.

Referring back to FIG. 1 , the input unit 130 may input an orthoimage 1 and a numerical surface model related to the target area by a user. Also, the input unit 130 may receive a user input by a user. Through this, the controller 150 may process a command corresponding to the user input.

The controller 150 may control overall operations of the open-air compost detection system 100 according to the present invention. For example, the controller 150 may input the orthoimage 1 to a pre-learned deep learning model to obtain a region of interest corresponding to the inputted orthoimage. In addition, the controller 150 may calculate current status information 2 for the region of interest detected in the target region.

The output unit 170 may output a result of a command processed according to a user input. For example, the output unit 170 generates the region of interest of the orthophoto 1 and current status information 2 related to the field compost in the region of interest through the control unit 150, the generated region of interest and current status Information (2) can be output.

Based on the configuration of the open-field compost detection system 100 described above, the method for detecting open-field compost will be described in more detail below.

3 is a flowchart of a method for detecting compost in an open field according to the present invention. 4 is a flowchart of a deep learning model learning method for learning a deep learning model prepared in a field compost detection system. 5 illustrates an embodiment of a region of interest obtained using a deep learning model prepared in a field compost detection system. 6 shows a plurality of modules provided to calculate current status information in the yard compost detection system. 7-16 illustrate one embodiment using a yard compost detection system.

Referring to FIG. 3, the open field compost detection system 100 according to the present invention acquires an ortho image 1 related to a target area (S100), inputs the ortho image 1 to a pre-trained deep learning model, In the orthoimage 1, a region of interest, which is a region where compost is present in the open field, may be obtained (S200).

To this end, the deep learning model learning system 200 may train a deep learning model using a plurality of ortho images 11 for learning and a plurality of regions of interest 12 for correct answers.

Specifically, referring to FIG. 4 , the deep learning model learning system 200 acquires a plurality of ortho images 11 for learning (S700), and as label data for each of the plurality of ortho images 11 for learning, a plurality of correct answer interests. Area 12 may be acquired (S800). Accordingly, the deep learning model learning system 200, when the orthoimage 1 is input to the deep learning model, outputs a region of interest 20, which is an area where compost in the field exists, in the orthoimage 1, so as to output a plurality of A deep learning model may be trained using the orthoimage 11 for learning and the region of interest 12 with a plurality of correct answers (S900).

At this time, the correct answer region of interest 12 may include a first region of interest that is not covered with a cover and a second region of interest that is a region that is covered with a cover among regions where open compost exists.

Through this, when the orthoimage 1 is input to the deep learning model, the deep learning model learning system 200 selects a first region of interest, which is an uncovered region, among regions in which the field compost exists in the orthoimage 1. The deep learning model may be trained so that the second region of interest, which is the region covered by the cover, is output.

Alternatively, when the orthoimage 1 is input to the deep learning model, the deep learning model learning system 200 selects a first region of interest, which is an uncovered region, among regions of interest in the orthoimage 1 where compost in the field exists. A deep learning model may be trained so that information related to the presence or absence of is output together with the region of interest.

Through the above configurations, the open field compost detection system 100 according to the present invention can obtain a region of interest corresponding to the orthophoto 1 by using a pre-learned deep learning model. At this time, the open-air compost detection system 100 may acquire the region of interest in various forms.

For example, the region of interest may be coordinate information about a region determined to be compost in an open field in the orthoimage 1 . In this case, the coordinate information may be information related to the location of a pixel in the orthoimage 1 or information related to a real location corresponding to the orthoimage 1 (eg, longitude, latitude).

For another example, the region of interest may be set such that a preset marker is generated in a region detected as field compost in the orthophoto 1 . In this case, the marker may be created to include a range corresponding to the area detected as open-field compost, or may be created at one point among the areas detected as open-field compost. In addition, the marker may be created to include information related to the compost in the open field (eg, presence or absence of a cover, area, etc.).

Through this, the field compost detection system 100 may calculate the location of the field compost through the position of the marker generated on the orthoimage 1, and may also be used to calculate the area, load, etc. of the field compost. there is.

Furthermore, the open-field compost detection system 100 according to the present invention includes a first region of interest, which is an uncovered region, among regions in which the open-field compost exists in the orthoimage 1, through a pre-learned deep learning model; At least one of the second regions of interest, which are regions covered by the cover, may be acquired.

Specifically, the region of interest may include a first region of interest related to a first color and a second region of interest related to a second color so as to determine whether or not the compost in the open field is covered. In this case, the first color is a color corresponding to the open compost and may include colors such as ocher, brown, reddish brown, earth color, and the like, and the second color is a color corresponding to the cover provided to cover the open compost, It may include colors such as blue, white, and green.

Accordingly, the deep learning model learning system 200 may train the deep learning model to detect the first ROI and the second ROI according to the first color and the second color.

In this regard, the region of interest may be set to include a first region of interest and a second region of interest based on the shape, material, and pattern of the compost and the cover in addition to the color.

Through this, the yard compost detection system 100 uses a pre-learned deep learning model to detect at least one of a first region of interest corresponding to a first color and a second region of interest corresponding to a second color in the orthoimage 1. you can get one.

Referring to FIG. 5 , for example, the open field compost detection system 100 may obtain a first region of interest 31 based on a first color corresponding to open field compost in the orthoimage 1 . In addition, the open compost detection system 100 may acquire the second region of interest 32 based on the second color corresponding to the cover of the open compost in the orthophoto 1 .

As another example, the open field compost detection system 100 may detect a region of interest corresponding to the field compost in the orthoimage 1 and determine whether a first color corresponding to the field compost is present in the detected region of interest. .

That is, the open compost detection system 100 determines that the first color is not covered if the first color is present in the region of interest, and if the first color is not present in the region of interest, the cover is covered. It can be determined as the second region of interest 32 that exists.

Referring back to FIG. 3 , the open field compost detection system 100 according to the present invention determines the priority for managing the open field compost based on at least one of the orthographic image 1 and the region of interest. Current status information (2) may be calculated, and the calculated current status information may be output.

Specifically, the open field compost detection system 100 may calculate various information related to open field compost as current status information 2 by applying various preset algorithms to the region of interest obtained in correspondence with the orthoimage 1. .

For example, the yard compost detection system 100 includes an area calculation module 111, a loading amount calculation module 112, a center coordinate calculation module 113, a water system distance calculation module 114, a designated location verification module 115, and A management state determination module 116 may be included. In this case, the current status information may include at least one of area, load capacity, location, water system distance, designated location, and management status.

Accordingly, the open compost detection system 100 may calculate the area of the region of interest using the area calculation module 111 . Specifically, the open compost detection system 100 may calculate the area of the open compost based on the number of pixels corresponding to the region of interest and the pixel area corresponding to any one of the plurality of pixels. .

That is, the open compost detection system 100 may calculate the area by multiplying the number of pixels corresponding to the region of interest by the pixel area of each pixel.

In this case, the number of pixels corresponding to the region of interest may be the number of pixels set as the region of interest in the orthoimage 1 . Also, the pixel area corresponding to any one pixel may be the area of the ground surface corresponding to one pixel of the orthoimage 1 . For example, pixel area may be spatial resolution.

Referring to FIG. 7 , for example, the open compost detection system 100 calculates the area of the open compost based on the pixel area 40 of a plurality of pixels detected as the region of interest 30 in the orthographic image 1. can do. In this case, the pixel area 40 may be calculated based on an actual distance corresponding to the horizontal length 42 of each pixel and an actual distance corresponding to the vertical length 41 of each pixel.

In addition, the open compost detection system 100 may calculate the loading amount of open compost for a region of interest using the loading calculation module 112 . Specifically, the open field compost detection system 100 obtains a numerical surface model corresponding to the region of interest, and for each of a plurality of pixels corresponding to the region of interest, the pixel area corresponding to each pixel and the numerical surface model in the numerical surface model. Based on the altitude (or height) of each pixel, the loading amount for the yard compost can be calculated.

That is, the field compost detection system 100 multiplies the pixel area corresponding to each pixel by the height of each pixel in the numerical surface model for each of a plurality of pixels corresponding to the region of interest, and for each of the plurality of pixels By summing up the calculated results, the load capacity for field compost can be calculated.

In this regard, the field compost detection system 100 multiplies the height of any one pixel among a plurality of pixels corresponding to the region of interest by the pixel area, and multiplies the resultant value by the number of the plurality of pixels to determine the field compost. load can be calculated. In this case, any one of the plurality of pixels may be a pixel having the highest height or a pixel having an average height among the plurality of pixels.

Referring to FIG. 8 , for example, the compost detection system 100 detects the amount of compost in the open field based on the pixel area 52 and height 51 of each of the plurality of pixels corresponding to the region of interest 30 . can be calculated In this case, the open field compost detection system 100 may extract an altitude 51 of each of a plurality of pixels corresponding to the region of interest 30 from the numerical surface model 50 .

Also, the open-air compost detection system 100 may determine the center coordinates of the region of interest using the center coordinate calculation module 113 . Specifically, the open-air compost detection system 100 may determine a coordinate corresponding to an arbitrary pixel within the ROI as the center coordinate. For example, the open compost detection system 100 may calculate the center of gravity of the region of interest and determine coordinates of pixels corresponding to the center of gravity as the center coordinates.

Referring to FIG. 9 , for example, the yard compost detection system 100 may determine the center coordinates 34 by calculating the center of gravity of the figure 33 formed to surround the outside of the region of interest 30. .

Also, the open compost detection system 100 may calculate the distance between the ROI and the water system detected in the orthoimage 1 as the water system distance using the water system distance calculation module 114 . Specifically, the open field compost detection system 100 detects a water system from the orthoimage 1, calculates the distance between the center coordinates determined for the region of interest and the shortest line among lines connecting one point of the detected water system, , the distance between the field compost and the water system can be calculated as the water system distance.

At this time, the open field compost detection system 100 may detect the water system in the orthographic image 1 using various techniques. For example, the field compost detection system 100 may determine the position of a water system in the orthographic image 1 based on a user input. For another example, the field compost detection system 100 may acquire the location of a water system by inputting the orthoimage 1 to a deep learning model prepared in advance to detect a water system. As another example, the open-field compost detection system 100 may determine a location at a lower altitude than a preset value as a water system based on a numerical surface model.

Referring to FIG. 10 , for example, the open field compost detection system 100 obtains the position of the water system 3 in the orthoimage 1, and the coordinates of the center of each region of interest 30a, 30b (or the region of interest). one side) and the shortest distance between the water system 3 can be calculated as the water system distance.

In addition, the open compost detection system 100 may verify whether the location of the region of interest exists at a pre-specified location using the specified location verification module 115 . Specifically, the open-air compost detection system 100 detects a designated area (eg, state-owned land or private land) set to determine whether the location where the open-air compost is stored is appropriate from the orthoimage 1, and determines the area of interest. It is possible to determine whether the center coordinates determined for belong to the designated area.

In this case, the open field compost detection system 100 may detect a designated area in the orthoimage 1 by using various techniques. For example, the open field compost detection system 100 may determine a designated area in the orthophoto 1 based on a user input. For another example, the open field compost detection system 100 may acquire a specified area by inputting the orthoimage 1 to a deep learning model prepared in advance to detect the specified area.

Referring to FIG. 11 , for example, the open field compost detection system 100 detects a state-owned area 4 as a designated area in an orthoimage 1, and each region of interest 30c and 30d overlaps the state-owned area. It can be determined whether or not it exists in an area. Accordingly, the open field compost detection system 100 may determine an area of interest 30c overlapping the state land area as an illegal land storage area, and may determine an area of interest 30d that does not overlap the state land area as a suitable land storage area.

For another example, the open-field compost detection system 100 detects the private land area 5 as a designated area in the orthoimage 1, and determines whether each of the regions of interest 30c and 30d overlaps the private land area. can determine whether Accordingly, the open field compost detection system 100 may determine an area of interest 30c that does not overlap with a private land area as an illegal landfill site, and may determine an area of interest 30d that overlaps with a private land area as an appropriate site area.

The open compost detection system 100 may use the management state determination module 116 to determine the management status according to the presence or absence of a cover for the open compost appearing in the region of interest. Specifically, when the ROI includes the first ROI and the second ROI, the compost detection system 100 acquires the first ROI and the second ROI corresponding to the orthoimage 1, respectively. It is possible to determine the management status of open-air compost based on whether

Referring to FIG. 12 , for example, when only the first regions of interest 30h and 30i are acquired corresponding to the orthophoto 1, the open compost detection system 100 detects the first regions of interest 30h and 30i. It is possible to determine a management state for the first management state (eg, “insufficient”). In addition, when only the second region of interest 30e is obtained corresponding to the orthoimage 1, the system 100 for detecting compost in the open field sets the management state of the second region of interest 30e higher than the first management state. It may be determined as the second management state (eg, “good”) set to indicate the level.

Furthermore, when the first region of interest and the second region of interest are obtained corresponding to the orthoimage 1, the open compost detection system 100 determines whether the first region of interest and the second region of interest are separated, It may be determined whether the obtained first and second ROI are one ROI, and based on the determination result, a management state of at least one of the first ROI and the second ROI may be determined.

For example, when the first ROI and the second ROI are combined, the open compost detection system 100 determines the first ROI and the second ROI as one third ROI 30g and 30h. and determine the management status of the third ROI as the first management status (eg, “insufficient”).

For another example, when the first ROI and the second ROI are separated, the open compost detection system 100 sets the management status for the first ROI 30h and 30i to the first management state (for example, For example, “insufficient”), and the management status of the second ROI 30e and 30f is set to a second management status (eg, “good”) set to indicate a higher level than the first management status. can decide

On the other hand, the open compost detection system 100 detects the region of interest based on the existence or nonexistence of the first region of interest when information related to the presence or absence of the first region of interest is obtained together with the region of interest in correspondence to the orthoimage 1. You can also judge the management status for .

For example, the open compost detection system 100 may determine the first management status (eg, "insufficient") as the management status for the ROI 30g, 30h, 30i, and 30j in which the first ROI exists. can

For another example, the open-air compost detection system 100 sets the management status for the ROI 30e and 30f where the first ROI does not exist to a level higher than the first management status, and sets the second management status ( For example, "good").

Through the configurations described above, the open field compost detection system 100 according to the present invention can detect the area where the open field compost exists in the orthographic image 1, and furthermore, the open field compost detection system 100 can detect the detected open field compost. By calculating the information related to the compost, the management state of the compost in the yard and the possibility of contamination by the compost in the yard can be predicted.

Furthermore, the open-air compost detection system 100 according to the present invention can output the current state information calculated above so that the user can check it.

Referring to FIG. 13 , for example, the open compost detection system 100 may generate at least one of image-type information, detection result-related information, monitoring-related information, and report-type information as current status information.

To this end, the open compost detection system 100 includes an image open input button 62, a detection result data input button 63, a monitoring data input button 64, and a report generation input button 65 through a display screen 60. ), and based on user input, current status information corresponding to each button can be generated (or output).

At this time, the open compost detection system 100 may output at least a part of information in the form of an image along with each of the above buttons through the display screen 60 to be shown to the user in advance.

Accordingly, when a user input corresponding to the open image input button 62 is received, the system 100 for detecting open compost may output an orthoimage 1 displaying an area where the detected compost in the open storage exists.

Referring to FIG. 14, the open-field compost detection system 100 creates a highlight mark (eg, a circular mark) on the areas 30k, 30l, and 30m where the open-field compost is detected in the orthographic image 1, It can be output to the display screen.

In addition, when a user input corresponding to the detection result data input button 63 is received, the open compost detection system 100 performs at least one of the calculated area, loading amount, center coordinates, water system distance, designated location, and management status. can output

At this time, the yard compost detection system 100 converts and outputs at least one of the calculated area, loading amount, center coordinates, water system distance, designated location, and management status in a previously prepared form (eg, tabular form). can

Referring to FIG. 15, the open compost detection system 100 calculates the area and loading amount in a pre-prepared form (eg, table form) based on a user input corresponding to the detection result data input button 63. (or volume), center coordinates (or longitude, latitude), water system distance, designated location (or national land), and management status (or status) can be output.

Also, when a user input corresponding to the monitoring data input button 64 is received, the system 100 for detecting compost in the field may output information related to monitoring of the compost in the field.

Here, the information related to monitoring of the compost in the open field may include information related to a result of detecting the compost in the open field and information related to management of the detected compost in the open field.

Referring to FIG. 16, the field compost detection system 100 inspects the field compost in a previously prepared form (eg, table form) based on a user input corresponding to the monitoring data data input button 64. Information related to schedule, management level, guidance order date according to management level, follow-up inspection schedule, and guidance completion can be output.

Here, the inspection schedule may include date information of detecting information related to open field compost through the orthoimage 1 . For example, the inspection schedule may include date information on a drone image, an aerial image, a satellite image, and the like to generate the orthoimage 1 .

The management level is a level set according to the management status of open compost, and may be set based on current status information calculated for a region of interest. That is, the open compost detection system 100 may determine a management level based on at least one of an area, a loading amount, a water system distance, a designated location, and a management state of the open compost.

For example, the open-air compost detection system 100 indicates the management level in a plurality of stages, but when the water system distance exceeds a preset distance (or when the ratio of the loading amount to the water system distance exceeds a preset level) E) It is lowered by one level, and the management level is lowered by one level when the area of interest is determined to be an illegal yard based on whether or not the designated location is located, and the management level can be determined to be lowered by one level when the management status is “insufficient”.

The guidance order date may include date information on a request for management of the open-field compost from a person (eg, a land owner or a local head of a local organization) related to an area where the open-field compost exists. In this case, the guidance command date may be set based on a user input.

Alternatively, when the management level of the open-field compost is lower than the preset level, the system 100 for detecting open-field compost may transmit a guidance command message for the open-field compost to a server device designated in advance in relation to the region where the corresponding open-field compost is located. there is. In this case, the open compost detection system 100 may set the time point at which the guidance command message is transmitted as the guidance command date. In this case, the guidance command message may include information related to compost in an open field and content requesting management of compost in an open field.

The follow-up inspection schedule may include information on a date set to check the continuous management status of the compost in the yard. That is, the follow-up inspection schedule may be set when re-detection of the compost in the open field is required after the detection of the compost in the open field according to the above inspection schedule.

For example, the open-air compost detection system 100 may set a date when a preset period elapses from the inspection schedule as a follow-up inspection schedule.

As another example, the field compost detection system 100 may set a follow-up inspection schedule when a guidance command date is set, and in this case, the field compost detection system 100 sets a predetermined period from the guidance command date. One date can be set as the follow-up inspection schedule.

Whether or not the system is completed may be set to indicate whether the management level is improved according to the detection result of the compost in the yard performed on different dates.

For example, the system 100 detecting compost in an open field according to a follow-up inspection schedule detects compost in an open field and determines a management level, the management level determined according to the inspection schedule and the management level determined according to the follow-up inspection schedule. , and according to the comparison result, if the management level determined according to the follow-up inspection schedule is higher than the management level determined according to the inspection schedule, it may be set to indicate that the guidance is completed.

For another example, the field compost detection system 100 determines the management level of the past field compost and the detected field compost when the field compost is detected within a preset distance range from the previously detected location of the field compost. Management grades for open-air compost are compared, and based on the comparison result, whether or not to complete the system can be set.

As another example, the open-air compost detection system 100 may set the guidance to be completed when the management level determined according to the follow-up inspection schedule is higher than the preset level.

In addition, when a user input corresponding to the report generation input button 65 is received, the open compost detection system 100 may generate and output a report including current status information.

Here, the report may be written in various formats, and the open compost detection system 100 may generate the report to include at least one of image-type information, detection result-related information, and monitoring-related information.

Through the configurations described above, the open-field compost detection system 100 according to the present invention outputs the current status of open-field compost detected from the orthophoto 1 in various forms, and furthermore, generates and outputs monitoring information on the open-field compost. By doing so, the user can be configured to easily manage a plurality of yard composts existing in various locations.

In addition, the open-field compost detection system 100 may provide information on open-field compost requiring priority management to the user by determining the inspection status of the open-field compost and management level of the open-field compost management state.

Furthermore, the present invention described above can be implemented as computer readable codes or instructions in a medium on which a program is recorded. That is, various control methods according to the present invention may be integrated or individually provided in the form of a program.

On the other hand, the computer-readable medium includes all types of recording devices in which data readable by a computer system is stored. Examples of computer-readable media include Hard Disk Drive (HDD), Solid State Disk (SSD), Silicon Disk Drive (SDD), ROM, RAM, CD-ROM, magnetic tape, floppy disk, optical data storage device, etc. there is

Furthermore, the computer-readable medium may be a server or cloud storage that includes storage and can be accessed by electronic devices through communication. In this case, the computer may download the program according to the present invention from a server or cloud storage through wired or wireless communication.

Furthermore, in the present invention, the above-described computer is an electronic device equipped with a processor, that is, a CPU (Central Processing Unit), and there is no particular limitation on its type.

On the other hand, the above detailed description should not be construed as limiting in all respects and should be considered as illustrative. The scope of the present invention should be determined by reasonable interpretation of the appended claims, and all changes within the equivalent scope of the present invention are included in the scope of the present invention.

1: Orthoscopic image
2: Status information
3: water system
4: state-owned territory
5: private property area
11: Orthoscopic image for learning
12: Answer area of interest
20: area of interest
30: region of interest
31 first region of interest
32 second region of interest
33: A figure formed to surround the outside of the region of interest
34: center coordinates
40: pixel area
41: vertical length
42: horizontal length
50: numerical surface model
51: Altitude
52: pixel area
100: Yard compost detection system
200: deep learning model learning system

Claims (10)

obtaining an orthoimage associated with the target area;
inputting the orthoimage to a pre-learned deep learning model, and acquiring a region of interest, which is an area where compost in the field exists, in the orthoimage;
calculating current status information related to the compost in the open field to determine a priority for managing the compost in the open field based on at least one of the orthoimage and the region of interest;
determining a management level for the region of interest based on the calculated status information; and
When the management level is lower than a preset level, transmitting a guidance command message for the open-field compost to a server designated in advance in relation to the region of interest in which the open-field compost exists;
In the step of obtaining the region of interest,
Obtaining at least one of a first region of interest corresponding to the open compost and a second region of interest corresponding to the cover, based on whether or not a cover is provided to prevent loss of the open compost;
In the step of determining the management level,
When only the first region of interest is acquired from the orthoimage, determining the management level as a first management level;
When only the second region of interest is acquired from the orthoimage, determining the management level as a second management level higher than the first management level;
When both the first region of interest and the second region of interest are acquired from the orthoimage, the management level is determined based on whether the first region of interest and the second region of interest are separated. detection method.
delete delete The method of claim 1, wherein the step of calculating current state information related to the open compost comprises:
calculating an area of the open compost based on the number of pixels corresponding to the region of interest and a pixel area corresponding to any one of the plurality of pixels; .
According to claim 1,
Acquiring a numerical surface model corresponding to the region of interest; further comprising,
In the step of calculating current status information related to the open compost,
For each of the plurality of pixels corresponding to the region of interest, calculating a loading amount for the pile compost based on the area corresponding to each pixel and the height of each pixel in the numerical surface model; Compost detection method.
According to claim 1,
Further comprising: detecting a water system in the orthoimage;
In the step of calculating current status information related to the open compost,
determining coordinates corresponding to one arbitrary pixel in the region of interest as center coordinates; and
Calculating the distance between the field compost and the water system as a water system distance by calculating the distance between the center coordinates determined for the region of interest and the shortest line among lines connecting the detected point in the water system; , Field compost detection method.
According to claim 1,
In the orthoimage, detecting a designated area set to determine whether the location where the yard compost is stored is suitable; further comprising;
In the step of calculating current status information related to the open compost,
determining coordinates corresponding to one arbitrary pixel in the region of interest as center coordinates; and
and determining whether the center coordinates determined for the region of interest belong to the designated region.
According to claim 1,
The pre-learned deep learning model,
It is learned according to the deep learning model learning method,
The deep learning model learning method,
Acquiring a plurality of orthoimages for learning;
acquiring a plurality of correct answer regions of interest as label data for each of the plurality of orthoimages for learning; and
When an orthoimage is input to the deep learning model, the deep learning model is learned using the plurality of orthoimages for learning and the plurality of answer regions of interest so that a region of interest, which is an area where compost in the field exists, is output in the orthoimage. A method for detecting yard compost, comprising the step of:
a storage unit for storing orthoimages related to the target area; and
By inputting the orthoimage to a pre-learned deep learning model, a region of interest, which is an area where compost in the field exists, is obtained in the orthoimage, and the compost in the field is determined based on at least one of the orthoimage and the region of interest. In order to determine the priority for management, current state information related to the open compost is calculated, and based on the calculated current state information, a management level for the area of interest is determined, and the management level is lower than a preset level. , a control unit for transmitting a guidance command message for the yard compost to a server designated in advance in relation to the region of interest;
The control unit,
Obtaining at least one of a first region of interest corresponding to the open compost and a second region of interest corresponding to the cover, based on whether or not a cover is provided to prevent loss of the open compost;
When only the first region of interest is acquired from the orthoimage, determining the management level as a first management level;
When only the second region of interest is acquired from the orthoimage, determining the management level as a second management level higher than the first management level;
When both the first ROI and the second ROI are acquired from the orthoimage, the management class is determined based on whether the first ROI and the second ROI are separated or not. yard compost detection system.
A program that is executed by one or more processes in an electronic device and stored in a computer-readable recording medium,
said program,
obtaining an orthoimage associated with the target region;
inputting the orthoimage to a pre-learned deep learning model, and acquiring a region of interest, which is an area where compost in the field exists, in the orthoimage;
calculating current status information related to the compost in the open field to determine a priority for managing the compost in the open field based on at least one of the orthoimage and the region of interest;
determining a management level for the region of interest based on the calculated status information; and
When the management level is lower than a preset level, transmitting a guidance command message for the open-field compost to a server designated in advance in relation to the region of interest in which the open-field compost exists;
In the step of obtaining the region of interest,
Obtaining at least one of a first region of interest corresponding to the open compost and a second region of interest corresponding to the cover, based on whether or not a cover is provided to prevent loss of the open compost;
In the step of determining the management level,
When only the first region of interest is acquired from the orthoimage, determining the management level as a first management level;
When only the second region of interest is obtained from the orthoimage, determining the management level as a second management level higher than the first management level;
When both the first ROI and the second ROI are obtained from the orthoimage, the management class is determined based on whether the first ROI and the second ROI are separated. A program stored on a recording medium that can be read by a computer.