Highway greening engineering inspection method and system based on artificial intelligence AI

Abstract

The invention relates to a highway greening engineering inspection method and system based on artificial intelligence AI. The system comprises a mobile device, a camera and a controller, wherein the camera and the controller are arranged on the mobile device; the controller is used for receiving the patrolling digital image, shot by the camera, of the greening project; research and development of artificial intelligence AI identification software and machine learning algorithm software of highway greening engineering diseases; performing analysis processing on the inspection digital image by artificial intelligence AI identification software and machine learning algorithm software of greening engineering diseases, and determining an abnormal area in the greening engineering; the mobile device is a vehicle or an unmanned aerial vehicle. And rapid movement can be realized. The inspection digital image is analyzed by adopting artificial intelligence AI identification software and machine learning algorithm software of greening engineering diseases, the types and areas of the diseases in the greening engineering are determined, and compared with manual naked eye judgment and identification in the prior art, the inspection efficiency of the highway greening engineering and the accuracy of disease identification are remarkably improved.

Description

Technical Field

[0001] The present disclosure relates to the technical field of highways, and in particular to an inspection method and system for highway greening projects based on artificial intelligence (AI). Background Art

[0002] The greening project of expressways plays an important role in beautifying the traffic environment, relieving fatigue of drivers and passengers, and preventing the night lights of oncoming vehicles from affecting the driver's vision. Due to the high speed and large traffic volume of expressways, once the driver's vision is affected, it will inevitably affect driving behavior and endanger traffic safety. Therefore, timely, scientific and reasonable maintenance is urgently needed. Timely and rapid discovery of diseases or hidden dangers and timely maintenance can effectively ensure the traffic safety of expressways and the safety of people's lives and property.

[0003] In the relevant technologies of highway inspection, manual inspection of tree defects is required, including, for example, whether trees are missing, dead, or whether the tree crown diameter exceeds the limit. These problems need to be judged by human eyes, which is inefficient, inaccurate, and also poses a great safety risk. Summary of the invention

[0004] In order to solve the problems existing in the current related technologies, the present disclosure provides a highway greening project inspection method and system based on artificial intelligence (AI).

[0005] According to a first aspect of an embodiment of the present disclosure, there is provided an inspection system for highway greening projects based on artificial intelligence AI, including: a mobile device, a camera, a controller, greening project disease artificial intelligence AI recognition software and machine learning algorithm software;

[0006] The camera and the controller are arranged on the mobile device; the artificial intelligence AI recognition software for greening engineering diseases and the machine learning algorithm software are arranged in the controller;

[0007] The mobile device is a vehicle or a drone;

[0008] The controller is used to control the mobile device to move along the target road, and adjust the movement trajectory and movement direction of the mobile device in real time according to the detection results, so as to obtain the digital image of the greening project comprehensively and clearly, and create favorable conditions for accurately identifying the diseases of the highway greening project;

[0009] The camera is used to digitally photograph the greening project on the target road to obtain a digital inspection image of the greening project;

[0010] The controller is also used to receive the inspection digital image of the greening project taken by the camera;

[0011] The artificial intelligence AI recognition software and machine learning algorithm software for greening project diseases are used to analyze the inspection digital image to determine the type and area of ​​the disease in the greening project, wherein the machine learning algorithm software provides a judgment basis for the artificial intelligence AI recognition software for greening project diseases to determine the type and area of ​​the disease in the greening project;

[0012] The greening project includes: greening projects on the central dividing strip and the roadside.

[0013] In the second aspect, the present application proposes a highway greening project inspection method based on artificial intelligence AI, which is applied to a highway greening project inspection system based on artificial intelligence AI, and the highway greening project inspection system based on artificial intelligence AI includes: a mobile device, a camera, a controller, greening project disease artificial intelligence AI recognition software and machine learning algorithm software; the camera and the controller are set on the mobile device; the mobile device is a vehicle or a drone; the greening project disease artificial intelligence AI recognition software and the machine learning algorithm software are set in the controller 2;

[0014] The method comprises: the controller controls the mobile device to move along the target road, and adjusts the movement track and movement direction of the mobile device in real time according to the detection result to perform more targeted detection;

[0015] The camera photographs the greening project on the target road to obtain a digital inspection image of the greening project;

[0016] The controller receives the inspection digital image of the greening project taken by the camera;

[0017] The artificial intelligence AI recognition software and machine learning algorithm software for greening project diseases are used to analyze the inspection digital image to determine the type and area of ​​the disease in the greening project, wherein the machine learning algorithm software provides a judgment basis for the artificial intelligence AI recognition software for greening project diseases to determine the type and area of ​​the disease in the greening project;

[0018] The greening project includes: greening projects in the central dividing strip of expressways and on the sides of roads.

[0019] The technical solution provided by the embodiments of the present disclosure may include the following beneficial effects: The inspection method and system of highway greening projects based on artificial intelligence AI of the present application can be completed by mobile devices, which can be vehicles or drones, instead of manual labor, and is very suitable for the characteristics of long mileage and heavy inspection tasks of highways. By setting a camera on a mobile device (including vehicles and drones), the camera digitally photographs the greening project on the target road to obtain a digital inspection image of the greening project. The controller receives the digital inspection image of the greening project taken by the camera; the digital inspection image is analyzed by the artificial intelligence AI recognition software for digital images of greening project diseases and the machine learning algorithm software to determine the abnormal areas in the greening project, which significantly improves the efficiency and accuracy of determining the abnormal areas in the greening project.

[0020] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present disclosure and, together with the description, serve to explain the principles of the present disclosure.

[0022] Figure 1 is a schematic diagram of a highway greening project inspection system according to an exemplary embodiment;

[0023] Figure 2 is a schematic diagram of tree loss according to an exemplary embodiment;

[0024] Figure 3 is a schematic diagram of a tree marking frame according to an exemplary embodiment;

[0025] Figure 4 is a schematic diagram of a tree dying according to an exemplary embodiment;

[0026] Figure 5 is a schematic diagram of a digital image frame of adjacent trees according to an exemplary embodiment;

[0027] Figure 6 is a schematic diagram of a digital image frame of adjacent trees according to an exemplary embodiment;

[0028] Figure 7 The figure is a schematic diagram of a digital image frame of adjacent trees according to an exemplary embodiment. DETAILED DESCRIPTION

[0029] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, the same numbers in different drawings represent the same or similar elements unless otherwise indicated. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present disclosure. Instead, they are merely examples of devices and methods consistent with some aspects of the present disclosure as detailed in the appended claims.

[0030] This application proposes a highway greening project inspection system 1 based on artificial intelligence (AI), see the attached Figure 1 The structure diagram of a highway greening project inspection system 1 based on artificial intelligence AI is shown; the system includes: a mobile device 4, a camera 3 and a controller 2, and artificial intelligence AI recognition software and machine learning algorithm software for greening project diseases; the camera 3 and the controller 2 are arranged on the mobile device 4. The artificial intelligence AI recognition software and machine learning algorithm software for greening project diseases are arranged in the controller 2.

[0031] The controller 2 includes a processor 21 and a memory 22 .

[0032] Specifically, the memory 22 may store the above-mentioned independently developed artificial intelligence AI recognition software and machine learning algorithm software for greening project diseases.

[0033] The processor 21 can run the above-mentioned artificial intelligence AI recognition software and machine learning algorithm software for greening project diseases.

[0034] The mobile device 4 is a vehicle or a drone.

[0035] The controller 2 is used to control the mobile device 4 to move along the target road, and adjust the movement track and movement direction of the mobile device 4 in real time according to the detection result to obtain more targeted detection.

[0036] In this embodiment, the processor 21 of the controller 2 includes a motion control module 211; the motion control module 211 is used to control the mobile device 4 to move along the target road, and adjust the motion track and motion direction of the mobile device 4 in real time according to the detection results to perform more targeted detection. For example, the motion control module 211 can control the specific movement of a vehicle or a drone.

[0037] The camera 3 is used to photograph the greening project on the target road and obtain the inspection digital image of the greening project.

[0038] The controller 2 is used to receive the inspection digital image of the greening project taken by the camera, and use the artificial intelligence AI recognition software for greening project diseases and the machine learning algorithm software to analyze the inspection digital image to determine the type and area of ​​the disease in the greening project.

[0039] The machine learning algorithm software provides a judgment basis for the artificial intelligence AI recognition software of the greening project diseases to determine the type and area of ​​the diseases in the greening project;

[0040] In this embodiment, the value of each parameter in the greening project is large and variable. By using a machine learning algorithm, a standard value can be determined from the massive parameter data, and the standard value is used as a basis for judgment. For example, determine the standard value of a size. A machine learning algorithm is used to study a large amount of size data to determine the standard value of the size. With a reliable standard value, it can be used as a reference. This application uses a machine learning algorithm to determine the standard value, which improves the accuracy of the standard value.

[0041] The artificial intelligence AI recognition software for greening project diseases is used to automatically identify greening project images, determine the measurement value, and compare the measurement value with the standard value stored in the machine learning algorithm software to determine whether there is a disease. For example, by determining the measurement value of a size and comparing the measurement value with the standard value, it is possible to determine whether the size is normal.

[0042] Specifically, the greening project inspection module 212 is used to analyze the inspection digital image to determine the type and area of ​​the disease in the greening project.

[0043] The sending module 213 is used to send the alarm information to the service platform.

[0044] The greening project includes: greening projects on the central dividing strip and the roadside.

[0045] The above-mentioned technical solution of the present application is that in the inspection system of the highway greening project, a mobile device is provided, and the mobile device is a vehicle or a drone. It can move quickly. The mobile device is also provided with a camera and a controller. The controller receives the inspection digital image of the greening project taken by the camera. The greening project disease artificial intelligence AI recognition software and machine learning algorithm software are used to analyze the inspection digital image to determine the type and area of ​​the disease in the greening project. The efficiency and accuracy of the inspection of the highway greening project are improved. Compared with the prior art, the inspection efficiency and accuracy are significantly improved.

[0046] In some embodiments, the controller is also used to identify and measure the tree spacing, geometric dimensions, color and chromaticity of the greening project based on the inspection digital images of the greening project and using the above-mentioned artificial intelligence AI recognition software for greening project diseases to determine the geometric dimensions, color and chromaticity of each target tree, as well as the tree spacing between any two adjacent target trees.

[0047] Determine abnormal areas in the central median and the roadside greening project where tree spacing or geometric dimensions do not meet the standards and specifications;

[0048] According to the geometric size and color of each target tree and the distance between any two adjacent target trees, it is determined whether each target tree is dead, infested or damaged.

[0049] The geometric dimensions may include one or more of the following: tree crown diameter, breast diameter, width, height, and dimensions of any length in any direction.

[0050] In this embodiment, the tree defect sample picture set includes pictures of missing trees, pictures of incomplete trees, pictures of dead trees, and pictures of trees intruding into the limit. After a large number of picture training, a machine learning algorithm for greening project defects can be obtained, which can store defects such as dead trees, incomplete trees, missing trees, and tree intrusion. The above-mentioned greening project disease artificial intelligence AI recognition software and the above-mentioned machine learning algorithm software can be used to identify the color, spacing and size of the target trees.

[0051] Determine abnormal areas in the central median and the roadside greening project where the tree spacing or height does not meet the standards and specifications.

[0052] See attached Figure 2 , tree 1, tree 2, tree 3 and tree 4, among which there is a missing tree between tree 2 and tree 3. Since the missing tree will directly reflect the spacing, if the spacing between two adjacent target trees is abnormally large, it can be proved that there is a missing tree between the two adjacent target trees. Therefore, the missing tree situation can be determined by identifying the spacing.

[0053] The above-mentioned AI recognition software for greening engineering diseases can be used to identify the color and chromaticity of trees, and compared with the color and chromaticity of trees in normal conditions. If there is a large difference, it means that the trees are dead. Another method is to use a machine learning algorithm to learn and record the color and chromaticity of dead trees, and compare and analyze with the color and chromaticity of trees identified by the above-mentioned AI recognition software for greening engineering diseases to identify whether the trees are dead, as shown in the attached Figure 4 shown.

[0054] The above-mentioned AI recognition software for greening engineering diseases can be used to calculate the tree spacing by using the horizontal pixels between the center lines of any two adjacent trees; similarly, the tree height can be measured by using the vertical pixels of any tree. By analyzing whether the spacing of the target trees is the standard spacing or height, the purpose of identifying the missing trees, whether the height is qualified, and whether the trees are dead can be achieved.

[0055] See attached Figure 3 , in some embodiments, in the marked box of each tree, determining the coordinates of the center point of the marked box;

[0056] The horizontal coordinate of the center point of the rectangular frame: X = (x1 + x2) / 2;

[0057] The vertical coordinate of the center point of the rectangular frame: Y = (y1 + y2) / 2;

[0058] Width of the rectangle: W = abs(x1-x2);

[0059] Height of the rectangular box: H = abs(y1-y2);

[0060] The height of the rectangle is used as the height of each tree. The width of the rectangle is used as the width of each tree.

[0061] The distance L1 between the two trees can be determined according to the horizontal coordinate values ​​of any two adjacent center points, where L1=X1-X2, wherein X1 and X2 are the horizontal coordinates of the center points of two adjacent rectangular frames, respectively.

[0062] Each adjacent distance Li can be calculated, where i represents the distance between the i+1th tree and the i-th data.

[0063] The average distance between any two trees can be calculated based on the above Li:

[0064]

[0065] n is a positive number greater than 0. The total number of trees is n+1.

[0066] It can be determined whether the distance between any two trees meets the standard requirements. Specifically, it can be determined whether the difference between the distance between any two trees and the standard distance is less than or equal to a predetermined difference threshold. If it is less than the predetermined difference threshold, it is determined that the distance between the two trees meets the standard requirements. If the difference between the distance between any two trees and the standard distance is greater than the predetermined difference threshold, it is determined that the distance between the two trees does not meet the standard requirements.

[0067] |Li-Lb|<∈;

[0068] Among them, ∈ is the difference threshold.

[0069] In one embodiment, if the second tree between the first tree and the third tree is missing, the distance L1 between the first tree and the third tree will be much greater than the standard distance Lb. In this case, it can be determined that replanting is required between the first tree and the third tree.

[0070] Specifically, the number of trees that need to be replanted can be calculated using the following formula:

[0071]

[0072] According to the distance between any two adjacent target trees, if it is greater than the standard distance, it can be determined whether there is a tree defect.

[0073] In some embodiments, in order to improve the accuracy of defect type identification results, the controller is also used to enhance the tree digital image training samples in the greening project artificial artificial intelligence AI recognition software to improve the balanced recognition ability and accuracy in the greening project artificial artificial intelligence AI recognition software.

[0074] Among the trees in the greening project of the highway, the most common defect is the death of trees. In some sections of the highway, a large number of trees will die due to the drought. The main reason for the loss of trees is that the trees died due to traffic accidents, or poor soil quality or drought, and the trees were not replaced in time. This disease is relatively rare. Among the training samples, a large number of sample digital images are the defect of death. If there are too many sample images of a certain defect, the artificial intelligence recognition software AI is prone to deviation in the location of the defect type when training. Therefore, it is necessary to enhance the pictures of this type of tree loss. For example, in a sample set, there are 500 pictures with the defect type of death; there are 50 pictures with the defect type of tree loss. In order to improve the balanced recognition ability of the model, it is necessary to expand the number of sample pictures with the defect type of tree loss from 50 to 500, so that the balanced recognition ability of the artificial intelligence recognition software AI can be significantly improved.

[0075] There are many methods to achieve sample enhancement for this defect type of missing trees. For example, blur processing technology can be used to add mosaic operations to the standard sample image once or multiple times to shield one or several trees in the green belt in the middle of the highway. In this way, data enhancement images of sample images with one or more missing trees can be obtained.

[0076] In order to realize the recognition of dead trees, the dead samples can be enhanced in the samples. The processing method can be to use a color transformation operation to perform one or more color transformation operations on the defective sample images, for example, transforming green into yellow, so as to obtain data enhanced images of one or more defective sample images of dead trees.

[0077] In some embodiments, the controller 2 is further configured to determine that there is a defect between any two adjacent target trees in response to a difference between the distance between the two adjacent target trees and the standard distance being greater than a predetermined difference threshold.

[0078] In some embodiments, the controller 2 is further configured to determine the distance between the two adjacent target trees by using the following method;

[0079] Step 1: Collect digital images of green trees in the central median of a highway, calculate the average spacing of green trees under normal conditions using the above-mentioned artificial intelligence AI recognition software for greening engineering diseases, and use the above-mentioned machine learning algorithm software to record and store them as the evaluation standard for the spacing of green trees in the central median of the highway;

[0080] Step 2: Taking the standard length L of the central divider guardrail as a reference, the greening project disease artificial intelligence AI recognition software is used to extract the number of pixels N1 of the digital photographic digital image of the central divider guardrail of the standard length;

[0081] Step 3: Using the inspection digital image of the greening project, draw parallel lines perpendicular to the driving direction with the trunks of the two adjacent target trees as the center;

[0082] Step 4: Use the above-mentioned artificial intelligence AI recognition software for greening engineering diseases to extract the number of pixels N2 between the two parallel lines;

[0083] Step 5: Determine the gap Gap between the two adjacent target trees according to the following formula:

[0084] Gap = N2 / N1×L.

[0085] In some embodiments, the controller 2 is further configured to determine that the target tree needs to be pruned in response to the difference between the geometric size of any target tree and the standard geometric size being greater than a predetermined size difference threshold, so as to avoid the occurrence of traffic accidents caused by the growth of tree branches beyond the limit;

[0086] Send an alarm message to the central control station, wherein the alarm message includes disease information and location information of the disease.

[0087] In this embodiment, if the height of the target tree determined by the above-mentioned AI recognition software for diseases in greening projects is too high or the diameter at breast height is too large in the digital image of the current target tree, the current target tree needs to be pruned. In actual projects, the excessive size of the tree at breast height may cause the tree to occupy the lane due to the excessive diameter at breast height, which is very likely to cause traffic accidents. For this reason, highway maintenance needs to prune trees with excessive diameter at breast height in a timely manner.

[0088] In some embodiments, the controller 2 is also used to, for any target tree, use the above-mentioned greening project disease artificial intelligence AI recognition software to determine the color and chromaticity of the leaves of the target tree, and determine the similarity by comparing with the color and chromaticity of standard leaves stored in the machine learning algorithm; if the similarity is less than or equal to the similarity threshold, it is determined that the target tree is dead.

[0089] In some embodiments, the color similarity between the color of the leaves of the target tree and the color of the standard target tree can be determined. If the color similarity is greater than a predetermined color similarity threshold, the target tree is determined to be dead.

[0090] In some embodiments, the chromaticity similarity between the chromaticity of the leaves of the target tree and the chromaticity of the standard target tree can be determined. If the chromaticity similarity is greater than a predetermined chromaticity similarity threshold, the target tree is determined to be dead.

[0091] In some embodiments, the two parameters of color and chromaticity can be used to jointly determine whether the target tree is dead. That is, in the embodiments including the above two cases, the target tree is determined to be dead only when the color meets the conditions and the chromaticity also meets the conditions. Specifically, the color similarity between the color of the leaves of the target tree and the color of the standard target tree is determined. If the color similarity is greater than a predetermined color similarity threshold, the chromaticity similarity difference between the chromaticity of the leaves of the target tree and the chromaticity of the standard target tree is determined. If the chromaticity similarity is greater than a predetermined chromaticity similarity threshold, the target tree is determined to be dead.

[0092] See attached Figure 5 In some embodiments, for any two adjacent digital image frames, feature points in each digital image frame may be obtained.

[0093] In the attached Figure 5In the figure, two adjacent digital image frames are shown. As shown in the figure, the same tree is located in these two digital image frames, so the two digital image frames need to be seamlessly spliced ​​for more accurate measurement. When performing seamless splicing, first, it is necessary to determine the feature points, which are the basis for splicing. Specifically, the scale-invariant feature transform (SIFT) algorithm can be used to extract the feature points of each partial area digital image frame. As shown in the figure, the feature point in the first digital image frame is A, and the feature point in the second digital image frame is A1. These two points are located in the two digital image frames, which are actually the same point on the target tree. The feature points can be used to align the two digital image frames according to the above two feature points.

[0094] The two adjacent digital image frames are spliced ​​and combined according to the feature points of each digital image frame to obtain a preliminary combined digital image frame to achieve horizontal alignment.

[0095] In some embodiments, the step of digital image cropping may be further included, where the overlapping parts of the digital images in the two digital image frames are cropped, that is, the digital image area on the right side of the feature point A in the first digital image and the area on the left side of the feature point A1 in the second digital image frame are overlapping areas and can be cropped and deleted.

[0096] See attached Figure 6 The schematic diagram of the preliminary combination of digital image frames shown is that after the preliminary combination, horizontal alignment is achieved.

[0097] In the preliminary combined digital image frame, determining tree edge features in the preliminary combined digital image frame;

[0098] The preliminary combined digital image is corrected in the vertical direction according to the tree edge features to obtain a corrected combined digital image.

[0099] In this embodiment, in the preliminary combined digital image frame, the Hough algorithm is used to determine the tree edge features of the target tree in the preliminary combined digital image frame.

[0100] Generally, the tree edge is a straight edge. Specifically, the straight edge is the straight edge where the feature point is located. The straight edge can be used to align the first digital image frame and the second digital image frame in the vertical direction.

[0101] The preliminary combined digital image is corrected according to the tree edge features to obtain a corrected combined digital image.

[0102] See attached Figure 7The corrected combined digital image is shown. The preliminary combined digital image is corrected in the vertical direction according to the straight line edge feature to achieve vertical alignment. After the above digital image processing, the two digital image frames can be seamlessly connected, which is closer to the real digital image and improves the authenticity of the combined digital image. For the corrected combined digital image, the digital image artificial intelligence recognition software AI is used to determine the size of the tree to obtain a more accurate size result.

[0103] In some embodiments, the controller 2 is also used to control the mobile device 4 to move along the target road, and adjust the movement trajectory and movement direction of the mobile device 4 in real time according to the detection results, so as to perform more targeted detection. Specifically, the following steps are included: controlling the movement speed of the mobile device according to the abnormal area in the greening project to perform more accurate measurement of the abnormal area in the greening project; specifically including: controlling the movement speed of the mobile device from fast to slow, controlling the mobile device to approach the abnormal area in the greening project; controlling the camera to shoot the abnormal area in the greening project again, so as to obtain a clearer inspection digital image of the abnormal area in the greening project; and further analyzing the clearer inspection digital image.

[0104] In some embodiments, after determining the abnormal area, the controller 2 is also used to send an alarm message to the central control console, and the alarm message includes specific location information. The specific location information may further include: the identification of the digital image frame, the location of the target tree in the digital image frame, so as to facilitate the subsequent maintenance work on the target tree. The platform can determine the distance from the starting point to the actual location of the target tree range photographed by the digital image frame through distance conversion based on the identification of the digital image frame. When converting the distance, the distance from the starting point to the actual location of the target tree range photographed by the digital image frame can be determined based on the actual distance corresponding to each digital image frame and the serial number of the digital image frame in the digital image frame sequence. Among them, the identification of the digital image frame includes the serial number of the digital image frame.

[0105] After receiving the above alarm information, the platform can send instructions to the maintenance vehicle to enable the maintenance personnel to rush to the target location to perform tree maintenance work, wherein the maintenance work includes: pruning the target trees and planting new trees in the vacant locations.

[0106] In some embodiments, the above-mentioned artificial intelligence AI recognition software for greening project diseases is developed to analyze and process digital photography of greening projects, and has the following characteristics:

[0107] Use digital camera pixels to calculate the distance between two adjacent green trees, and determine whether there are tree defects based on this;

[0108] Intelligently identify and measure the color, chromaticity and shape of greening trees, and compare and analyze them with the normal status of greening projects stored in machine learning algorithms to determine whether the trees are dead. Based on the changes in shape, determine whether the trees have dried up due to lack of water and need timely watering and maintenance.

[0109] In some embodiments, the machine learning algorithm software is developed to store the size, leaf color, chromaticity and shape of greening trees in normal conditions as a comparison standard for intelligent identification of greening project diseases, thereby determining whether the trees are dead, damaged, have branch infestations or are suffering from water shortage and drying diseases.

[0110] On the second aspect, the present application proposes a highway greening project inspection method based on artificial intelligence AI, which is applied to the above-mentioned highway greening project inspection system based on artificial intelligence AI. The highway greening project inspection system based on artificial intelligence AI includes: a mobile device, a camera, a controller, greening project disease artificial intelligence AI recognition software and machine learning algorithm software; the camera and controller are set on the above-mentioned mobile device; the above-mentioned greening project disease artificial intelligence AI recognition software and machine learning algorithm software are set in the above-mentioned controller 2.

[0111] The mobile device is a vehicle or a drone;

[0112] The method comprises: the controller controls the mobile device to move along the target road, and adjusts the movement track and movement direction of the mobile device in real time according to the detection result to perform more targeted detection;

[0113] The camera photographs the greening project on the target road to obtain a digital inspection image of the greening project;

[0114] The controller receives the inspection digital image of the greening project taken by the camera;

[0115] The artificial intelligence AI recognition software for greening project diseases and the machine learning algorithm software are used to analyze the inspection digital images to determine the type and area of ​​the diseases in the greening project, wherein the machine learning algorithm software provides a judgment basis for the artificial intelligence AI recognition software for greening project diseases to determine the type and area of ​​the diseases in the greening project;

[0116] The greening project includes: greening projects in the central dividing strip of expressways and on the sides of roads.

[0117] In some embodiments, the method may further include the following steps:

[0118] The controller uses artificial intelligence AI recognition software for greening project diseases to identify the tree spacing, geometric dimensions and color of the greening project based on the inspection digital image of the greening project, and determines the geometric dimensions, color and chromaticity of each target tree, as well as the tree spacing between any two adjacent target trees;

[0119] Determine abnormal areas in the central median and the roadside greening project where tree spacing or geometric dimensions do not meet the standards and specifications;

[0120] According to the color, geometric size of each target tree and the distance between any two adjacent target trees, it is determined whether each target tree is dead, intruded or damaged.

[0121] In some embodiments, the method may further include the following steps:

[0122] In response to the difference between the distance between any two adjacent target trees and the standard distance being greater than a predetermined difference threshold, the controller determines that there is a defect between the two adjacent target trees.

[0123] In some embodiments, the method may further include the following steps:

[0124] The controller determines the distance between the two adjacent target trees by the following method;

[0125] Step 1: Collect digital images of green trees in the central median of the expressway, identify and calculate the average spacing of green trees in a normal state through the greening engineering disease artificial intelligence AI recognition software, and use the machine learning algorithm software to record and store them as the evaluation standard for the spacing of green trees in the central median of the expressway;

[0126] Step 2: Taking the standard length L of the central dividing strip guardrail as a reference, the greening project artificial intelligence AI recognition software is used to extract the number of pixels N1 of the digital photographic digital image of the central dividing strip guardrail of the standard length;

[0127] Step 3: Using the inspection digital image of the greening project, draw parallel lines perpendicular to the driving direction with the trunks of the two adjacent target trees as the center;

[0128] Step 4: Use the artificial intelligence AI recognition software for greening engineering diseases to extract the number of pixels N2 between the two parallel lines;

[0129] Step 5: Determine the gap Gap between the two adjacent target trees according to the following formula:

[0130] Gap = N2 / N1×L.

[0131] In some embodiments, the method may further include the following steps:

[0132] In response to the difference between the geometric dimensions of any target tree and the standard geometric dimensions being greater than a predetermined difference threshold, the controller determines that the target tree needs to be pruned to prevent the tree branches from growing beyond the limit and causing traffic accidents;

[0133] Send an alarm message to the central control station, wherein the alarm message includes disease information and location information of the disease.

[0134] In some embodiments, the method may further include the following steps:

[0135] The controller uses the artificial intelligence AI recognition software for greening engineering diseases to identify the color and chromaticity of the leaves of any target tree, determines the color and chromaticity of the leaves of the target tree, and compares the color and chromaticity with the color and chromaticity of the standard leaves stored in the machine learning algorithm software;

[0136] If the similarity is less than or equal to the similarity threshold, it is determined that the target tree is dead.

[0137] In some embodiments, the method may further include the following steps:

[0138] For any two adjacent digital image frames, obtain feature points in each digital image frame;

[0139] splicing and combining the two adjacent digital image frames according to the feature points of each digital image frame to obtain a preliminary combined digital image frame;

[0140] In the preliminary combined digital image frame, determining tree edge features in the preliminary combined digital image frame;

[0141] The preliminary combined digital image is corrected in the vertical direction according to the straight line edge feature to obtain a corrected combined digital image.

[0142] In some embodiments, the method may further include the following steps:

[0143] The controller performs enhanced processing on the tree digital image training samples in the greening project disease artificial intelligence AI recognition software. Through digital image training, a machine learning algorithm can be generated and optimized to improve the balanced recognition ability and accuracy in the greening project disease artificial intelligence AI recognition software.

[0144] Those skilled in the art will readily appreciate other embodiments of the present disclosure after considering the specification and practicing the present disclosure. This application is intended to cover any variations, uses or adaptations of the present disclosure, which follow the general principles of the present disclosure and include common knowledge or customary technical means in the art that are not disclosed in the present disclosure. The specification and examples are to be considered as exemplary only, and the true scope and spirit of the present disclosure are indicated by the following claims.

[0145] It should be understood that the present disclosure is not limited to the exact structures that have been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present disclosure is limited only by the appended claims.

Claims

1. A highway greening project inspection system based on artificial intelligence AI, characterized in that: include: Mobile devices, cameras, controllers, artificial intelligence AI recognition software for greening project diseases, and machine learning algorithm software; The mobile device is a vehicle or a drone; the camera and the controller are arranged on the mobile device; The artificial intelligence AI recognition software for greening engineering diseases and the machine learning algorithm software are arranged in the controller; The controller is used to control the mobile device to move along the target road, and adjust the movement trajectory and movement direction of the mobile device in real time according to the detection result, so as to perform more targeted detection; The camera is used to photograph the greening project on the target road and obtain the inspection digital image of the greening project; The controller is also used to receive the inspection digital image of the greening project taken by the camera; The artificial intelligence AI recognition software for greening project diseases and the machine learning algorithm software are used to analyze the inspection digital images to determine the type and area of ​​the disease in the greening project. The machine learning algorithm software provides a judgment basis for the artificial intelligence AI recognition software of the greening project diseases to determine the type and area of ​​the diseases in the greening project; The greening project includes: greening projects in the central dividing strip of expressways and on the sides of roads.

2. The highway greening project inspection system based on artificial intelligence AI according to claim 1 is characterized in that: The controller is also used to identify the tree spacing, geometric dimensions, color and chromaticity of the greening project based on the inspection digital image of the greening project and using artificial intelligence AI recognition software for greening project diseases, and determine the geometric dimensions, color and chromaticity of each target tree, as well as the tree spacing between any two adjacent target trees; Determine abnormal areas in the central median and the roadside greening project where tree spacing or geometric dimensions do not meet the standards and specifications; According to the color, geometric size of each target tree and the distance between any two adjacent target trees, it is determined whether each target tree is dead, intruded or damaged.

3. The highway greening project inspection system based on artificial intelligence AI according to claim 2 is characterized in that: The controller is further configured to determine that there is a defect between any two adjacent target trees in response to a difference between the distance between the two adjacent target trees and the standard distance being greater than a predetermined difference threshold.

4. The highway greening project inspection system based on artificial intelligence AI according to claim 3 is characterized in that: The controller is also used to determine the distance between the two adjacent target trees by adopting the following method; Step 1: Collect digital images of green trees in the central median of the expressway, identify and calculate the average spacing of green trees in a normal state through the artificial intelligence AI recognition software for greening engineering diseases, and use the machine learning algorithm software to record and store them as an evaluation standard for the spacing of green trees in the central median of the expressway; Step 2: Taking the standard length L of the central dividing strip guardrail as a reference, the artificial intelligence AI recognition software for greening engineering diseases is used to extract the number of pixels N1 of the digital photographic digital image of the central dividing strip guardrail of the standard length; Step 3: Using the inspection digital image of the greening project, draw parallel lines perpendicular to the driving direction with the trunks of the two adjacent target trees as the center; Step 4: Using the artificial intelligence AI recognition software for greening engineering diseases to extract the number of pixels N2 between the two parallel lines; Step 5: Determine the gap Gap between the two adjacent target trees according to the following formula: Gap = N2 / N1×L.

5. The highway greening project inspection system based on artificial intelligence AI according to claim 1 is characterized in that: The controller is also used to, in response to a difference between the geometric dimensions of any target tree and the standard geometric dimensions being greater than a predetermined difference threshold, determine that the target tree needs to be pruned to avoid the occurrence of traffic accidents caused by the growth of tree branches encroaching on the limit; Send an alarm message to the central control station, wherein the alarm message includes disease information and location information of the disease.

6. The highway greening project inspection system based on artificial intelligence AI according to claim 1 is characterized in that: The controller is also used to, for any target tree, identify the color and chromaticity of the leaves of the target tree using the artificial intelligence AI recognition software for greening engineering diseases, determine the color and chromaticity of the leaves of the target tree, and determine the similarity by comparing the color and chromaticity of the leaves with the standard color and chromaticity of the leaves stored in the machine learning algorithm; If the similarity is less than the similarity threshold, it is determined that the target tree is dead.

7. The highway greening project inspection system based on artificial intelligence AI according to claim 1 is characterized in that: For any two adjacent digital image frames, feature points in each digital image frame can be obtained; splicing and combining the two adjacent digital image frames according to the feature points of each digital image frame to obtain a preliminary combined digital image frame; In the preliminary combined digital image frame, determining tree edge features in the preliminary combined digital image frame; The preliminary combined digital image is corrected in the vertical direction according to the straight line edge feature to obtain a corrected combined digital image.

8. The highway greening project inspection system based on artificial intelligence AI according to claim 2 is characterized in that: The controller is also used to enhance the tree digital image training samples in the artificial intelligence AI recognition software for greening project diseases. Through digital image training, a machine learning algorithm can be generated and optimized to improve the balanced recognition ability and accuracy in the artificial intelligence AI recognition software for greening project diseases.

9. The highway greening project inspection system based on artificial intelligence AI according to claim 1 is characterized in that: The artificial intelligence AI recognition software for the greening project diseases is developed to analyze and process the digital photography of the greening project, and has the following characteristics: Use digital camera pixels to calculate the distance between two adjacent green trees, and determine whether there are tree defects based on this; Intelligently identify and measure the color, chromaticity and morphology of greening trees, and compare and analyze them with the normal state of greening projects stored by machine learning algorithms to determine whether the trees are dead. Based on the changes in morphology, determine whether the trees are dry due to lack of water and need to be watered and maintained in time; The machine learning algorithm software is developed to store the size, leaf color, chromaticity and shape of greening trees in normal conditions as a comparison standard for intelligent identification of greening project diseases, thereby determining whether the trees are dead, damaged, have branch infestations or are suffering from water shortage and drying diseases.

10. A highway greening project inspection method based on artificial intelligence AI, characterized in that: Applied to a highway greening project inspection system based on artificial intelligence AI, the highway greening project inspection system based on artificial intelligence AI includes: a mobile device, a camera, a controller, greening project disease artificial intelligence AI recognition software and machine learning algorithm software; the camera and the controller are arranged on the mobile device; the greening project disease artificial intelligence AI recognition software and the machine learning algorithm software are arranged in the controller; The mobile device is a vehicle or a drone; The method comprises: the controller controls the mobile device to move along the target road, and adjusts the movement track and movement direction of the mobile device in real time according to the detection result to perform more targeted detection; The camera photographs the greening project on the target road to obtain a digital inspection image of the greening project; The controller receives the inspection digital image of the greening project taken by the camera; The artificial intelligence AI recognition software for greening project diseases and the machine learning algorithm software are used to analyze and process the inspection digital images to determine the abnormal areas in the greening project; wherein the machine learning algorithm software provides a judgment basis for the artificial intelligence AI recognition software for greening project diseases to determine the type and area of ​​the diseases in the greening project; The greening project includes: greening projects in the central dividing strip of expressways and on the sides of roads.

Images