Agricultural land value evaluation optimization method, device and equipment and readable storage medium

[0064] It should be understood that the specific embodiments described here are only used to explain the application, but not to limit the application.

[0065] The embodiment of the application provides an optimization method for the evaluation of agricultural land value. In the first embodiment of the method for evaluation and optimization of the agricultural land value of the application, refer to figure 1 , The method for evaluating and optimizing the value of agricultural land includes:

[0066] Step S10, acquiring an image of the land to be assessed corresponding to the land to be assessed, and inputting the image of the land to be assessed into a preset image classification model, so as to classify the land image to be assessed to obtain an image classification result;

[0067] In this embodiment, it should be noted that the preset image classification model is a machine learning model that has been trained based on deep learning, and the land images to be assessed include satellite images, camera images, and the like.

[0068] Acquire the land image to be assessed corresponding to the land to be assessed, and input the land image to be assessed into a preset image classification model to classify the land image to be assessed to obtain the image classification result, specifically, through a preset shooting method Acquire an image of the land to be assessed corresponding to the land to be assessed, wherein the preset shooting mode includes shooting methods such as satellite shooting, drone shooting, and mobile phone shooting, and then input the to-be-assessed land image into a preset image classification model to Perform data processing on the pixel matrix to be evaluated corresponding to the land image to be evaluated based on the data processing layer in the preset image classification model, and output the image classification result, wherein the data processing layer includes a convolutional layer and a pool. The data processing includes convolution processing, pooling processing, and full connection, etc., wherein the convolution layer is used to perform the convolution processing, and the pooling layer is used to perform all In the pooling process, the fully connected layer is used to perform the fully connected.

[0069] Among them, it should be noted that the convolution processing process can be understood as: the statistical characteristics of one part of the image feature are the same as the other parts, that is, the statistical characteristics learned in this part can also appear in the corresponding other part, so it will learn The statistical characteristics obtained are used as detectors and applied to any place of this image feature, that is, the statistical characteristics learned through a small-scale image are convolved with the image characteristics of the original large-size image. In mathematics, convolution can be It is the characteristic matrix of the corresponding image multiplied by multiple detection matrices in advance and finally summed to obtain the convolution processing result. The pooling processing includes pooling processing including maximum pooling, average pooling, etc., specifically, first The result of the convolution processing is divided into a plurality of pixel matrices with a preset size. If the maximum value is pooled, the maximum pixel value of the pixel matrix is ​​used instead of the pixel matrix to obtain a new image matrix, that is, to obtain Pooling processing result, the full connection can be regarded as a special convolution processing, the result of the special convolution processing is to obtain a one-dimensional vector corresponding to the image, that is, the pixel matrix corresponding to the image is transformed through the full connection It is a one-dimensional vector, and the one-dimensional vector includes the combination information of all the features of the image corresponding to the one-dimensional vector.

[0070] Wherein, in step S10, the step of inputting the land image to be assessed into a preset image classification model to classify the land image to be assessed, and obtaining an image classification result includes:

[0071] Step S11, input the land image to be assessed into a preset image classification model, so as to perform a preset number of alternate processing of convolution and pooling on the land image to be assessed to obtain a convolution pooling processing result;

[0072] In this embodiment, the land image to be assessed corresponds to a unique image matrix to be assessed, wherein the image matrix to be assessed is a pixel matrix, and the pixel matrix is ​​composed of multiple pixel values.

[0073] Input the to-be-assessed land image into a preset image classification model to perform a preset number of alternate processing of convolution and pooling on the to-be-assessed land image to obtain the convolutional pooling processing result, specifically, the The assessed land image is input into a preset image classification model to perform convolution processing on the to-be-assessed image matrix corresponding to the land image to be assessed to obtain a convolution processing result, and then pooling the convolution processing result, The pooling processing result is obtained, and further, the convolution processing and the pooling processing are repeatedly alternately performed on the pooling processing result until the number of alternations of the convolution processing and the pooling processing reaches a preset number of times, then Obtain the convolution pooling processing result.

[0074] Step S12: Fully connect the convolution pooling processing result to obtain an image classification vector, and obtain the image classification result from the image classification vector.

[0075] In this embodiment, the convolutional pooling processing result is fully connected to obtain an image classification vector, and the image classification result is obtained from the image classification vector, specifically, the convolutional pooling processing The results are fully connected to convert the convolution processing result into a corresponding one-dimensional feature vector, where the one-dimensional feature vector is the image classification vector, and the one-dimensional feature vector includes one or more Feature coding, each of the feature codes is for a kind of semantics, and then an image classification feature code is extracted from the image classification vector, and then the image classification result is queried based on the image classification feature code.

[0076] Wherein, the land image to be assessed includes satellite images,

[0077] The step of obtaining the land image to be assessed corresponding to the land to be assessed, and inputting the land image to be assessed into a preset image classification model to classify the land image to be assessed, and the step of obtaining the image classification result includes:

[0078] Step A10, receiving the agricultural land coordinate information submitted by the target user corresponding to the land to be assessed, and determining the validity of the agricultural land coordinate information;

[0079] In this embodiment, the agricultural land coordinate information submitted by the target user corresponding to the land to be assessed is received, and the validity of the agricultural land coordinate information is determined. Specifically, the target user corresponding to the land to be assessed is received. Agricultural land coordinate information to obtain the longitude and latitude information of the plot in the agricultural land coordinate information, and the longitude and latitude information of the city where the plot is located, and further obtain the longitude and latitude information of the activity range corresponding to the activity scope of the loan user, and then based on the land The latitude and longitude information of the block, the latitude and longitude information of the city where the plot is located, and the latitude and longitude information of the activity range are used to determine the validity of the agricultural land coordinate information through a preset validity judgment equation group, wherein the preset validity judgment equation group As follows:

[0080] 0 <180

[0081] 0

[0082] (x-XX_x) 2 +(y-XX_y) 2 2

[0083] (x-YY_x) 2 +(y-YY_y) 2 2

[0084] Wherein, the longitude and latitude corresponding to the longitude and latitude information of the plot is (x, y), the longitude and latitude corresponding to the longitude and latitude information of the city where the plot is located is (XX_x, XX_y), and the longitude and latitude of the activity range longitude and latitude information is (YY_x, YY_y) , D 1 Is the preset first distance, d 2 Is a preset second distance, where the preset first distance is a distance used to ensure that the distance between the agricultural land and the city where the agricultural land is located is less than a preset first distance threshold, and the preset second distance The distance is a distance used to ensure that the distance between the agricultural land and the loan user is less than the preset second distance threshold.

[0085] Step A20, if the agricultural land coordinate information is valid, take the satellite image based on the agricultural land coordinate information;

[0086] In this embodiment, if the agricultural land coordinate information is valid, the position of the agricultural land is determined based on the agricultural land coordinate information, and the satellite image is captured by satellite photography.

[0087] Step A30, if the agricultural land coordinate information is invalid, return invalid information to the target user.

[0088] In this embodiment, if the agricultural land coordinate information is invalid, invalid information is returned to the target user. Specifically, if the agricultural land coordinate information is invalid, an error code is returned to the loan user to improve the The loan user resubmits the agricultural land coordinate information.

[0089] Step S20, based on the image classification result, determine the land evaluation value corresponding to the land to be evaluated;

[0090] In this embodiment, it should be noted that the land appraisal value can be used to formulate a loan plan for the target user, where the loan plan includes the number of loan installments, the loan amount range, etc., and the image classification result includes Classification results of wasteland, no planting for 5 years, no planting for 3 years, no planting for 1 year and continuous planting.

[0091] Based on the image classification result, determine the land evaluation value corresponding to the land to be assessed, specifically, based on the image classification result, select and determine the initial land evaluation value corresponding to the land to be assessed, and obtain the land to be assessed Corresponding to the credit score of the target user, wherein the credit score is related to the target user’s loan records, repayment records, personal assets and other customer information, and then the credit score is included in the land value evaluation reference factor, Then, based on the value evaluation factor and the initial land evaluation value, the land evaluation value is determined. For example, assuming that the image classification result is wasteland, the corresponding initial land value is determined to be 10000, and the credit score is obtained as If the score is 80, the credit score is included in the lending reference factor, and the value of the lending reference factor is 0.8, then the land appraisal value is 8000.

[0092] Wherein, the step of determining the land evaluation value corresponding to the land to be evaluated based on the image classification result includes:

[0093] Step S21: Determine the initial land evaluation value corresponding to the land to be evaluated based on the image classification result;

[0094] In this embodiment, based on the image classification result, the initial land evaluation value corresponding to the land to be evaluated is determined, specifically, based on the result code corresponding to the image classification result, the result code is selected from a preset database The corresponding initial land evaluation value, wherein the result code is an identifier of the image classification result, and the result code includes character strings, numbers, and the like.

[0095] Step S22: Obtain user data corresponding to the land to be assessed, and score the land to be assessed based on the user information, obtain a first scoring result, and include the first scoring result in a land value evaluation reference factor ;

[0096] In this embodiment, it should be noted that the user profile includes user information such as loan records, repayment records, and personal assets of the target user.

[0097] Acquire user data corresponding to the land to be assessed, and score the land to be assessed based on the user information, obtain a first scoring result, and include the first scoring result into the land value evaluation reference factor, specifically , Obtain the user information of the target user corresponding to the land to be assessed, and input the user information into a preset scoring model to score the target user, obtain the user scoring result, and use the user scoring result as the result The first scoring result of the land to be assessed is further included in the land value evaluation reference factor, so as to update the actual value of the land value evaluation reference factor.

[0098] Step S23: Determine the land evaluation value based on the land value evaluation reference factor and the initial land evaluation value.

[0099] In this embodiment, the land evaluation value is determined based on the land value evaluation reference factor and the initial land evaluation value. Specifically, the actual value is calculated based on the actual value of the land value evaluation reference factor. The product of the initial land evaluation value, and the product is used as the land evaluation value.

[0100] Step S30, periodically collecting images at each time point corresponding to the land to be assessed, and inputting the images at each time point into a preset image recognition model to obtain the land change status corresponding to the land to be assessed, and based on the land change Condition adjusts the estimated value of the land.

[0101] In this embodiment, it should be noted that the land change conditions include planting conditions and crop growth conditions, where the planting conditions include whether the land is planted with crops, etc., and the crop growth conditions include good growth, poor growth, etc. Growth status, the preset image recognition model is a machine learning model that has been trained based on deep learning.

[0102] Regularly collect images at each time point corresponding to the land to be assessed, and input the images at each time point into a preset image recognition model to obtain the land change status corresponding to the land to be assessed, and adjust the location based on the land change status. The land evaluation value, specifically, periodically collect images at each time point corresponding to the land to be assessed, wherein each time point image includes at least a first time point image and a second time point image, and each time The point image is input into a preset image recognition model to perform frame difference processing on the first time point image and the second time point image to obtain a difference matrix, and then input the difference matrix into the volume in the image recognition model The product neural network performs a preset number of convolution and pooling alternate processing on the difference matrix to obtain the convolution and pooling alternate processing results, and fully connects the convolution and pooling alternate processing results to obtain Image recognition vector, and then obtain each image recognition feature code in the image recognition, and then query the land change status based on each image recognition feature code, and then count the land change status into the land value evaluation reference factor, and Based on the land value evaluation reference factor, adjust the land evaluation value.

[0103] Wherein, in step S30, the step of adjusting the land evaluation value based on the land change status includes:

[0104] Step S31, scoring the land to be assessed based on the land change status to obtain a second scoring result;

[0105] In this embodiment, based on the land change status, the land to be assessed is scored to obtain a second score result, specifically, the land change status code corresponding to the land change status is obtained, and the land change status code is obtained according to the land change status. The second scoring result corresponding to the status code query, wherein the land change status code is an identifier of the land change status.

[0106] In step S32, the second scoring result is included in a land value evaluation reference factor, and the land evaluation value is adjusted according to the land value evaluation reference factor.

[0107] In this embodiment, the second scoring result is included in the land value evaluation reference factor, and the land evaluation value is adjusted according to the land value evaluation reference factor, specifically, the second scoring result is included Land value evaluation reference factor to update the land value evaluation reference factor, and then calculate the product of the updated land value evaluation reference factor and the land evaluation value to adjust the land evaluation value, for example, if the The second scoring result is 110, and the land value evaluation reference factor is 1, then the second scoring result is included in the land value evaluation reference factor, and the land value evaluation reference factor obtained is 1.1. If it is 10,000, the adjusted estimated value of the land is 11,000.

[0108] In this embodiment, an image of the land to be assessed corresponding to the land to be assessed is acquired, and the land image to be assessed is input into a preset image classification model to classify the land image to be assessed to obtain an image classification result, which is then based on the According to the result of image classification, the land evaluation value corresponding to the land to be assessed is determined, and the images at each time point corresponding to the land to be assessed are collected regularly, and the images at each time point are input into a preset image recognition model to obtain the land to be assessed. Evaluate the land change status corresponding to the land, and adjust the land evaluation value based on the land change status. That is, this embodiment obtains the land to be assessed image of the land to be assessed, and inputs the land to be assessed into a preset image classification model to classify the land to be assessed, obtain the image classification result, and obtain the image classification result according to the According to the image classification results, the land evaluation value of the land to be assessed can be determined, and then by regularly collecting images of the land to be assessed, and through a preset image recognition model, it can be determined whether the land to be assessed has corresponding crops and crops to be assessed. The planting status of the land and whether the target user can complete the planting of crops and other land change conditions within the expected time, so as to adjust the land evaluation value based on the land change status. That is, this embodiment provides a method for evaluating the value of agricultural land. The evaluation of the value of agricultural land can be realized through the preset image classification model and the preset image recognition model acquired based on deep learning. Furthermore, it realizes the accurate evaluation of agricultural land loans without the need for on-site inspections, thereby improving the efficiency of agricultural land value evaluation, so the technical problem of low efficiency of agricultural land value evaluation is solved.

[0109] Further, refer to figure 2 , Based on the first embodiment of the present application, in another embodiment of the present application, in step S30, each of the time point images includes a first time point image and a second time point image, and the preset image recognizes The model includes a convolutional neural network,

[0110] The step of inputting each of the time point images into a preset image recognition model to obtain the land change status corresponding to the land to be assessed includes:

[0111] Step A10, input each of the time-point images into a preset image recognition model to perform frame difference processing on the first-time-point image and the second-time-point image to obtain a difference matrix;

[0112] In this embodiment, it should be noted that the time point image includes at least a first time point image and a second time point image, and the interval between shooting the first time point image and shooting the second time point image Is a preset time period, and the preset time period can be set by the user. For example, if the preset time period is set to one month, the image taken on the first day of the preset time period is the first An image at a time point, and an image taken on the second day of the preset time period is the second time point image.

[0113] Input each of the time point images into a preset image recognition model to perform frame difference processing on the first time point image and the second time point image to obtain a difference matrix. Specifically, each of the time point images A preset image recognition model is input to perform frame difference processing on the pixel matrix corresponding to the first time point image and the pixel matrix corresponding to the second time point image to obtain the difference matrix.

[0114] Wherein, the step of performing frame difference processing on the first time point image and the second time point image to obtain a difference matrix includes:

[0115] Step A11, respectively acquiring a first pixel matrix corresponding to the first time point image and a second pixel matrix corresponding to the second time point image;

[0116] In this embodiment, the first pixel matrix corresponding to the first time point image and the second pixel matrix corresponding to the second time point image are obtained respectively, and specifically, the first time point image and the second pixel matrix are respectively obtained The second time point image is input into a preset image conversion model, and a first pixel matrix corresponding to the first time point image and a second pixel matrix corresponding to the second time point image are output, wherein the preset image conversion The model can be implemented based on MATLAB (matrix&laboratory, matrix factory).

[0117] Step A12: Perform a subtraction operation on the first pixel matrix and the second pixel matrix to obtain the difference matrix.

[0118] In this embodiment, the first pixel matrix and the second pixel matrix are subtracted to obtain the difference matrix. Specifically, the corresponding one of the first pixel matrix and the second pixel matrix is The pixel values ​​are subtracted to obtain the difference matrix.

[0119] Step A20: Input the difference matrix into the convolutional neural network to perform alternate processing of convolution and pooling on the difference matrix to obtain the land change status.

[0120] In this embodiment, the difference matrix is ​​input to the convolutional neural network to perform alternate processing of convolution and pooling on the difference matrix to obtain the land change status, specifically, the difference matrix is ​​input The convolutional neural network performs a preset number of alternate processing of convolution and pooling on the difference matrix to obtain convolution and pooling processing results, and then fully connecting the convolution and pooling processing results, Obtain the land change status vector, and then query the land change status based on the feature code in the land change status vector. For example, suppose the land change status vector is (1, 1), where 0 represents the agricultural land It is determined that the crop to be planted is corn, and 1 means that the corn is growing well.

[0121] In this embodiment, by inputting each of the time-point images into a preset image recognition model, frame difference processing is performed on the first-time-point image and the second-time-point image to obtain a difference matrix, and then the difference matrix Input the convolutional neural network to perform alternate processing of convolution and pooling on the difference matrix to obtain the land change status. That is, the present application collects agricultural land images at each time point and performs frame difference on the agricultural land images to obtain a difference matrix, and then input the difference matrix into a preset image recognition image to output the land change status In turn, it realizes the evaluation of the value of the agricultural land after the loan without conducting on-site inspections, reducing the loan risk, and avoiding the low efficiency of the agricultural land value evaluation due to the manual inspection after the loan. Therefore, it laid a foundation for solving the technical problem of low efficiency in agricultural land value evaluation.

[0122] Reference image 3 , image 3 It is a schematic diagram of the device structure of the hardware operating environment involved in the solution of the embodiment of the present application.

[0123] Such as image 3 As shown, the agricultural land value evaluation optimization device may include: a processor 1001, such as a CPU, a memory 1005, and a communication bus 1002. Among them, the communication bus 1002 is used to implement connection and communication between the processor 1001 and the memory 1005. The memory 1005 may be a high-speed RAM memory, or a stable memory (non-volatile memory), such as a magnetic disk memory. Optionally, the memory 1005 may also be a storage device independent of the aforementioned processor 1001.

[0124] Optionally, the agricultural land value evaluation and optimization equipment may also include a rectangular user interface, a network interface, a camera, an RF (Radio Frequency) circuit, a sensor, an audio circuit, a WiFi module, and so on. The rectangular user interface may include a display screen (Display) and an input sub-module such as a keyboard (Keyboard), and the optional rectangular user interface may also include a standard wired interface and a wireless interface. The network interface can optionally include a standard wired interface and a wireless interface (such as a WI-FI interface).

[0125] Those skilled in the art can understand, image 3 The structure of the agricultural land value evaluation and optimization equipment shown in does not constitute a limitation on the agricultural land value evaluation and optimization equipment, and may include more or less components than shown in the figure, or a combination of certain components, or different component arrangements.

[0126] Such as image 3 As shown, the memory 1005 as a computer storage medium may include an operating system, a network communication module, and an agricultural land value evaluation optimization program. The operating system is a program that manages and controls the hardware and software resources of the agricultural land value evaluation optimization equipment, and supports the operation of the agricultural land value evaluation optimization program and other software and/or programs. The network communication module is used to realize the communication between the components in the memory 1005 and the communication with other hardware and software in the agricultural land value evaluation and optimization system.

[0127] in image 3 In the illustrated agricultural land value evaluation optimization device, the processor 1001 is configured to execute the agricultural land value evaluation optimization program stored in the memory 1005 to implement the steps of the agricultural land value evaluation optimization method described in any one of the above.

[0128] The specific implementation of the agricultural land value evaluation optimization equipment of this application is basically the same as the foregoing embodiments of the agricultural land value evaluation optimization method, and will not be repeated here.

[0129] The embodiment of the application also provides an agricultural land value evaluation and optimization device, which is applied to agricultural land value evaluation and optimization equipment, and the agricultural land value evaluation and optimization device includes:

[0130] The image classification module is used to obtain the land image to be assessed corresponding to the land to be assessed, and input the land image to be assessed into a preset image classification model to classify the land image to be assessed and obtain an image classification result;

[0131] The loan module is used to determine the land evaluation value corresponding to the land to be assessed based on the image classification result;

[0132] The adjustment module is used to periodically collect images at each time point corresponding to the land to be assessed, and input the images at each time point into a preset image recognition model to obtain the land change status corresponding to the land to be assessed, and based on the The land change status adjusts the land evaluation value.

[0133] Optionally, the adjustment module includes:

[0134] A frame difference processing unit, configured to input each of the time point images into a preset image recognition model to perform frame difference processing on the first time point image and the second time point image to obtain a difference matrix;

[0135] The obtaining unit is configured to input the difference matrix into the convolutional neural network to perform alternate processing of convolution and pooling on the difference matrix to obtain the land change status.

[0136] Optionally, the frame difference processing unit includes:

[0137] An acquiring subunit, configured to respectively acquire a first pixel matrix corresponding to the first time point image and a second pixel matrix corresponding to the second time point image;

[0138] The subtraction operation unit is configured to perform a subtraction operation on the first pixel matrix and the second pixel matrix to obtain the difference matrix.

[0139] Optionally, the image classification module includes:

[0140] The convolution and pooling alternate processing unit is used to input the to-be-assessed land image into a preset image classification model to perform a preset number of convolution and pooling alternate processings on the to-be-assessed land image to obtain a convolution pool Chemical treatment result;

[0141] The fully connected unit is configured to fully connect the convolution pooling processing result to obtain an image classification vector, and obtain the image classification result from the image classification vector.

[0142] Optionally, the lending module includes:

[0143] A determining unit, configured to determine the initial land evaluation value corresponding to the land to be evaluated based on the image classification result;

[0144] The first scoring unit is used to obtain user information corresponding to the land to be assessed, and to score the land to be assessed based on the user information, obtain a first scoring result, and include the first scoring result in the land Value evaluation reference factor;

[0145] The lending unit is used to determine the land evaluation value based on the land value evaluation reference factor and the initial land evaluation value.

[0146] Optionally, the adjustment module further includes:

[0147] The second scoring unit is used to score the land to be assessed based on the land change status to obtain a second scoring result;

[0148] The adjustment unit is configured to include the second scoring result into the land value evaluation reference factor, and adjust the land evaluation value according to the land value evaluation reference factor.

[0149] Optionally, the agricultural land value evaluation and optimization device further includes:

[0150] The judgment module is used to receive the agricultural land coordinate information submitted by the target user corresponding to the land to be assessed, and determine the validity of the agricultural land coordinate information;

[0151] A photographing module, configured to photograph the satellite image based on the agricultural land coordinate information if the agricultural land coordinate information is valid;

[0152] An error reporting module is used to return invalid information to the target user if the agricultural land coordinate information is invalid.

[0153] The specific implementation of the agricultural land value evaluation optimization device of this application is basically the same as the above embodiments of the agricultural land value evaluation optimization method, and will not be repeated here.

[0154] The embodiment of the present application provides a readable storage medium, and the readable storage medium stores one or more programs, and the one or more programs may also be executed by one or more processors for implementation The steps of the method for evaluating and optimizing the value of agricultural land as described in any one of the above.

[0155] The specific implementation of the readable storage medium of the present application is basically the same as the foregoing embodiments of the agricultural land value evaluation optimization method, and will not be repeated here.

[0156] The above are only preferred embodiments of this application, and do not limit the scope of this application. Any equivalent structure or equivalent process transformation made using the content of the description and drawings of this application, or directly or indirectly used in other related technical fields , The same reason is included in the scope of patent processing of this application.