Road disease inspection method and system based on visual identification, terminal and storage medium

Abstract

The invention discloses a road disease inspection method and system based on visual identification, a terminal and a storage medium, and the method comprises the steps: obtaining the road video data of a target region, and carrying out the preprocessing of the road video data, and obtaining the target road video data; inputting the target road video data into a trained road data recognition model to obtain a road recognition result, and labeling the road recognition result to obtain road disease data; and filtering the road disease data to obtain target road disease data, and obtaining a road disease inspection result according to the target road disease data. According to the method, the road diseases can be intelligently recognized, meanwhile, the types of the road diseases can be intelligently classified, the inspection efficiency of road disease inspection and the accuracy of the inspection result are improved, and therefore the roads can be maintained in time.

Description

Technical Field

[0001] This invention relates to the field of road inspection technology, and in particular to a road defect inspection method, system, terminal, and computer-readable storage medium based on visual recognition. Background Technology

[0002] As road construction continues to expand, the importance of road maintenance is becoming increasingly prominent. Before carrying out road maintenance, it is necessary to inspect the existing roads for defects. Existing methods for road defect inspection mainly rely on manual operation and basic equipment. Inspectors drive around to inspect roads, and when problems are found, they get out of the vehicle to conduct on-site inspections and measurements. They then use mobile terminals (such as smartphones) to take photos, fill in road defect information, and report it. However, this method is not only cumbersome to operate, but also has limited inspection distance, resulting in low inspection efficiency. Furthermore, it suffers from the problem of strong subjectivity in human judgment, leading to low accuracy of inspection results.

[0003] Therefore, existing technologies still need to be improved and developed. Summary of the Invention

[0004] The main objective of this invention is to provide a road defect inspection method, system, terminal, and storage medium based on visual recognition, aiming to solve the problems of low inspection efficiency and low accuracy of inspection results in existing road defect inspection methods.

[0005] To achieve the above objectives, the present invention provides a road defect inspection method based on visual recognition, the road defect inspection method based on visual recognition comprising the following steps: Acquire road video data of the target area, and preprocess the road video data to obtain target road video data; The target road video data is input into a trained road data recognition model to obtain road recognition results, and the road recognition results are labeled to obtain road defect data. The road defect data is filtered to obtain target road defect data, and the road defect inspection results are obtained based on the target road defect data.

[0006] Optionally, the road defect inspection method based on visual recognition, wherein the step of acquiring road video data of the target area and preprocessing the road video data to obtain target road video data, further includes: Acquire a dataset of historical road images and create a data recognition neural network with a preset resolution; The resolution of the historical road image dataset is converted and adjusted according to the preset resolution to obtain the target historical road image dataset. The data recognition neural network is then trained based on the target historical road image dataset to obtain the road data recognition model.

[0007] Optionally, the road defect inspection method based on visual recognition, wherein acquiring road video data of the target area and preprocessing the road video data to obtain target road video data specifically includes: The target road in the target area is obtained, road video data of the target road is collected, and the data processing format and data processing resolution of the road data recognition model are obtained. The road video data is converted according to the data processing format to obtain road video data in a preset format. The resolution of the road video data in the preset format is adjusted according to the data processing resolution to obtain the target road video data.

[0008] Optionally, the road defect inspection method based on visual recognition, wherein the step of inputting the target road video data into a trained road data recognition model to obtain road recognition results, and then labeling the road recognition results to obtain road defect data, specifically includes: The recognition range is defined according to the data processing resolution of the road data recognition model to obtain the recognition calibration range, and the target road video data is input into the road data recognition model. The road data recognition model performs recognition processing on the target road video data according to the recognition calibration range to obtain road recognition results, wherein the road recognition results include the corner coordinates of the road image, the recognition type, and the type recognition accuracy; The identification type and the identification accuracy are labeled to obtain the road defect type and road defect specification, and road defect data is obtained based on the road defect type and the road defect specification.

[0009] Optionally, the road defect inspection method based on visual recognition, wherein the step of defining the recognition range according to the data processing resolution of the road data recognition model to obtain the recognition calibration range specifically includes: The data processing resolution of the road data recognition model is obtained. The horizontal pixel value of the data processing resolution is compared with the first preset data to obtain the upper margin. The vertical pixel value of the data processing resolution is subtracted from the second preset value to obtain the lower margin. Set a precision threshold for disease type identification accuracy, and generate a corresponding identification calibration range based on the precision threshold, the upper margin, and the lower margin.

[0010] Optionally, the road defect inspection method based on visual recognition, wherein the step of filtering the road defect data to obtain target road defect data, and obtaining road defect inspection results based on the target road defect data, specifically includes: The coordinates of the center coordinate point of the defined identification range are calculated to obtain the center coordinate point value, and corresponding data filtering conditions are formulated based on the center coordinate point value and the defined identification range. The road defect data is filtered according to the data filtering conditions to obtain target road defect data. The severity of the road defect is analyzed according to the road defect specifications of the target road defect data to obtain the severity of the hazard. The road defect inspection results are obtained based on the target road defect data and the severity of the hazard. The data filtering conditions are that the center coordinate point value is greater than the upper edge of the identification range, the center coordinate point value is less than the lower edge of the identification range, and the type identification accuracy is greater than the accuracy threshold of the identification range.

[0011] Optionally, the road defect inspection method based on visual recognition, wherein the step of filtering the road defect data to obtain target road defect data, and obtaining road defect inspection results based on the target road defect data, further includes: Obtain the inspection results of all roads in the target area, compile the inspection results of all roads into a table, obtain a road disease statistics table, and store the road disease statistics table in the road disease database.

[0012] Optionally, the road defect inspection method based on visual recognition, wherein the road defect inspection system based on visual recognition includes: The data processing module is used to acquire road video data of the target area, preprocess the road video data, and obtain target road video data. The data recognition module is used to input the target road video data into the trained road data recognition model to obtain the road recognition result, and to annotate the road recognition result to obtain road defect data; The road defect inspection module is used to filter the road defect data to obtain target road defect data, and to obtain road defect inspection results based on the target road defect data.

[0013] In addition, to achieve the above objectives, the present invention also provides a terminal, wherein the terminal includes: a memory, a processor, and a road defect inspection program based on vision recognition stored in the memory and executable on the processor, wherein when the road defect inspection program based on vision recognition is executed by the processor, it implements the steps of the road defect inspection method based on vision recognition as described above.

[0014] In addition, to achieve the above objectives, the present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium stores a road defect inspection program based on visual recognition, and the road defect inspection program based on visual recognition, when executed by a processor, implements the steps of the road defect inspection method based on visual recognition as described above.

[0015] In this invention, road video data of a target area is acquired, preprocessed to obtain target road video data, input into a trained road data recognition model to obtain road recognition results, and labeled to obtain road defect data. The road defect data is then filtered to obtain target road defect data, and road defect inspection results are derived based on this data. This invention can intelligently identify road defects and intelligently classify their types, improving the efficiency and accuracy of road defect inspections, thus enabling timely road maintenance. Attached Figure Description

[0016] Figure 1 This is a flowchart of a preferred embodiment of the road defect inspection method based on visual recognition of the present invention; Figure 2 This is a schematic diagram of the overall process of the road defect inspection method based on visual recognition of the present invention; Figure 3 This is a schematic diagram illustrating the presence of cracks in a road in a preferred embodiment of the present invention; Figure 4 This is a structural diagram of a preferred embodiment of the road defect inspection system based on visual recognition of the present invention; Figure 5 This is a structural diagram of a preferred embodiment of the terminal of the present invention. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of this invention clearer and more explicit, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0018] It should be noted that if the embodiments of the present invention involve directional indicators (such as up, down, left, right, front, back, etc.), the directional indicators are only used to explain the relative positional relationship and movement of the components in a certain specific posture (as shown in the figure). If the specific posture changes, the directional indicators will also change accordingly.

[0019] Furthermore, if the embodiments of this invention involve descriptions such as "first" or "second," these descriptions are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined with "first" or "second" may explicitly or implicitly include at least one of those features. Additionally, the technical solutions of the various embodiments can be combined with each other, but this must be based on the ability of those skilled in the art to implement them. If the combination of technical solutions is contradictory or impossible to implement, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection claimed by this invention.

[0020] The preferred embodiment of the road defect inspection method based on visual recognition described in this invention, such as... Figure 1 As shown, the road defect inspection method based on visual recognition includes the following steps: Step S10: Obtain road video data of the target area, and preprocess the road video data to obtain target road video data.

[0021] Specifically, in this embodiment of the invention, to address the problems of low inspection efficiency and low accuracy of existing road defect inspection methods, this invention proposes a road defect inspection method based on visual recognition. This method is executed through a terminal, which can perform real-time road defect inspection based on visual recognition. The corresponding terminal can be a DVR (Digital Video Recorder), a dashcam, or various devices with cameras. The specific implementation process of the visual recognition-based road defect inspection method is as follows: Figure 2As shown, firstly, before inspecting the corresponding roads for defects, a pre-trained road data recognition model needs to be embedded in the terminal to automatically execute the corresponding road defect recognition program. The training process of the road data recognition model is as follows: acquiring a historical road image dataset and creating a data recognition neural network with a preset resolution. Since the collected road image data or road video data has a corresponding resolution, this invention trains data recognition neural networks with different resolutions to improve the accuracy of recognition. The resolution of the historical road image dataset is converted and adjusted according to the preset resolution to obtain a target historical road image dataset. The data recognition neural network is then trained based on the target historical road image dataset to obtain the road data recognition model.

[0022] When the corresponding road defect identification program is executed, the required identification calibration range is set. This calibration range refers to the area designated for identifying road defects. The calibration range is determined by the road data identification model. Specifically, the data processing resolution of the road data identification model (e.g., 640*640) is obtained. The horizontal pixel value of the data processing resolution (640) is compared with a first preset value (e.g., 2) to obtain the upper margin (640 / 2). The vertical pixel value of the data processing resolution (640) is subtracted from a second preset value (e.g., 30) to obtain the lower margin (640-30). The corresponding identification calibration range is generated based on the upper and lower margins. The identification calibration range also includes a precision threshold for disease type identification accuracy (e.g., 0.6). The identification calibration range can be represented as: BaseBox(b_y1=640 / 2, b_y2=640-30, b_threshold=0.6), or BaseBox(b_y1=320, b_y2=610, b_threshold=0.6), where BaseBox is the identification calibration range, b_y1 is the top margin, b_y2 is the bottom margin, and b_threshold is the precision threshold. Next, it is necessary to determine whether the road data identification model has been successfully loaded. If it has not been successfully loaded, the terminal can be restarted. If it has been successfully loaded, the corresponding data acquisition command is generated.

[0023] The target road in the target area is determined according to the data acquisition instruction, and road video data of the target road is acquired, that is, road video data in YUV420, NV12, NV21, I420, YV12 formats, etc., is read from the camera. After acquiring the road video data, the format and resolution of the road video data need to be adjusted. Specifically, the data processing format and data processing resolution of the road data recognition model are obtained; the road video data is converted according to the data processing format to obtain road video data in a preset format; and the resolution of the road video data in the preset format is adjusted according to the data processing resolution to obtain the target road video data. In this embodiment of the invention, the road data recognition model trained has two resolutions: 720*720 and 640*640. Taking the 640*640 resolution road data recognition model as an example, the video format required for the 640*640 resolution road data recognition model is RGB24. Therefore, the road video data is adjusted to the target road video data in RGB24 format.

[0024] Step S20: Input the target road video data into the trained road data recognition model to obtain the road recognition result, and then label the road recognition result to obtain road defect data.

[0025] Specifically, after completing the format conversion of the road video data to obtain the target road video data, as follows: Figure 2 As shown, the target road video data needs to be imported into a road data recognition model. The road data recognition model processes the target road video data according to a pre-defined recognition calibration range to obtain a road recognition result. This result is a rectangular frame of the road image, including the corner coordinates of the road image (including the upper left and lower right corner coordinates), the recognition type, and the type recognition accuracy. It also includes the geographic latitude and longitude location information of the current device. The recognition types include potholes, round manhole covers, square manhole covers, cracks, road patches, transverse cracks, longitudinal cracks, broken slabs, landslides, spilled materials, and block cracks. Then, the recognition types and type recognition accuracy are labeled to obtain road defect types and road defect specifications. Based on these road defect types and specifications, road defect data is obtained, such as... Figure 3 As shown, Figure 3The text indicates that the current asphalt road has cracking defects; the road defect data can be represented as: Box(x1, y1, x2, y2, classType, threshold), where Box is the road defect data, x1 is the horizontal coordinate of the upper left corner of the rectangular frame, y1 is the vertical coordinate of the upper left corner of the rectangular frame, x2 is the horizontal coordinate of the lower right corner of the rectangular frame, y2 is the vertical coordinate of the lower right corner of the rectangular frame, classType is the recognition type, and threshold is the type recognition precision.

[0026] Step S30: Filter the road defect data to obtain target road defect data, and obtain road defect inspection results based on the target road defect data.

[0027] Specifically, after obtaining road defect data, such as Figure 2 As shown, the corresponding road defect data needs to be stored and reported. Before storage and reporting, the road defect data needs to be filtered. The reason for data filtering is that not all road defect data is needed, and road defect data with low accuracy thresholds need to be discarded. In this embodiment of the invention, data filtering is performed based on Box and BaseBox. Specifically, the coordinates of the center coordinate point of the identified range are calculated to obtain the center coordinate point value, and corresponding data filtering conditions are formulated based on the center coordinate point value and the identified range. The data filtering conditions are that the center coordinate point value is greater than the upper margin of the identified range, the center coordinate point value is less than the lower margin of the identified range, and the type identification accuracy is greater than the accuracy threshold of the identified range. The data filtering conditions can be expressed as: (centerY>b_y1&¢erY)<b_y2&&threshold> (b_threshold), where centerY is the center coordinate value; only road defect data that simultaneously meets the above three conditions is required, that is, the road defect data must be within the defined identification range; if the above three conditions cannot be met simultaneously, the next round of data identification is entered, and so on repeatedly. The road defect data is filtered according to the data filtering conditions to obtain target road defect data. The severity of the damage is analyzed based on the road defect specifications of the target road defect data to obtain the severity level. Finally, the road defect inspection results are obtained based on the target road defect data and the severity level.

[0028] Furthermore, the inspection results of road defects in the target area are obtained, and the inspection results of all roads are tabulated to obtain a road defect statistics table, which is then stored in the road defect database. The purpose of tabulating the inspection results of all roads is to facilitate the quick retrieval of the road defect inspection results for the corresponding road. For example, the user can input the inspection time, identification type, target area, and vehicle license plate number (any one of these can be entered), and the search results can be displayed in the table, thereby quickly extracting the road defect inspection results of the current road and promptly notifying relevant personnel to maintain the current road.

[0029] This invention can intelligently identify road defects and intelligently classify the types of road defects, thereby improving the inspection efficiency and accuracy of road defect inspections and enabling timely road maintenance.

[0030] Furthermore, such as Figure 4 As shown, based on the above-described road defect inspection method based on visual recognition, the present invention also provides a road defect inspection system based on visual recognition, wherein the road defect inspection system based on visual recognition includes: Data processing module 51 is used to acquire road video data of the target area, preprocess the road video data, and obtain target road video data; The data recognition module 52 is used to input the target road video data into the trained road data recognition model to obtain the road recognition result, and to perform annotation processing on the road recognition result to obtain road defect data; The road defect inspection module 53 is used to filter the road defect data to obtain target road defect data, and to obtain road defect inspection results based on the target road defect data.

[0031] Furthermore, such as Figure 5 As shown, based on the above-mentioned road defect inspection method based on visual recognition, the present invention also provides a terminal, which includes a processor 10, a memory 20 and a display 30. Figure 5 Only some of the terminal components are shown; however, it should be understood that it is not required to implement all of the components shown, and more or fewer components may be implemented instead.

[0032] In some embodiments, the memory 20 may be an internal storage unit of the terminal, such as a hard disk or memory. In other embodiments, the memory 20 may be an external storage device of the terminal, such as a plug-in hard disk, smart media card (SMC), secure digital card (SD), flash card, etc. Further, the memory 20 may include both internal and external storage devices. The memory 20 is used to store application software and various types of data installed on the terminal, such as the program code installed on the terminal. The memory 20 can also be used to temporarily store data that has been output or will be output. In one embodiment, the memory 20 stores a road defect inspection program 40 based on visual recognition, which can be executed by the processor 10 to implement the road defect inspection method based on visual recognition in this application.

[0033] In some embodiments, the processor 10 may be a central processing unit (CPU), a microprocessor, or other data processing chip, used to run program code stored in the memory 20 or process data, such as executing the vision-based road defect inspection method.

[0034] In some embodiments, the display 30 may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, or an OLED (Organic Light-Emitting Diode) touchscreen. The display 30 is used to display information on the terminal and to display a visual user interface.

[0035] In one embodiment, when the processor 10 executes the vision-based road defect inspection program 40 in the memory 20, the following steps are performed: Acquire road video data of the target area, and preprocess the road video data to obtain target road video data; The target road video data is input into a trained road data recognition model to obtain road recognition results, and the road recognition results are labeled to obtain road defect data. The road defect data is filtered to obtain target road defect data, and the road defect inspection results are obtained based on the target road defect data.

[0036] The step of acquiring road video data of the target area and preprocessing the road video data to obtain target road video data includes, prior to: Acquire a dataset of historical road images and create a data recognition neural network with a preset resolution; The resolution of the historical road image dataset is converted and adjusted according to the preset resolution to obtain the target historical road image dataset. The data recognition neural network is then trained based on the target historical road image dataset to obtain the road data recognition model.

[0037] Specifically, the step of acquiring road video data of the target area and preprocessing the road video data to obtain target road video data includes: The target road in the target area is obtained, road video data of the target road is collected, and the data processing format and data processing resolution of the road data recognition model are obtained. The road video data is converted according to the data processing format to obtain road video data in a preset format. The resolution of the road video data in the preset format is adjusted according to the data processing resolution to obtain the target road video data.

[0038] Specifically, the step of inputting the target road video data into a trained road data recognition model to obtain road recognition results, and then labeling the road recognition results to obtain road defect data, includes: The recognition range is defined according to the data processing resolution of the road data recognition model to obtain the recognition calibration range, and the target road video data is input into the road data recognition model. The road data recognition model performs recognition processing on the target road video data according to the recognition calibration range to obtain road recognition results, wherein the road recognition results include the corner coordinates of the road image, the recognition type, and the type recognition accuracy; The identification type and the identification accuracy are labeled to obtain the road defect type and road defect specification, and road defect data is obtained based on the road defect type and the road defect specification.

[0039] Specifically, the step of defining the recognition range based on the data processing resolution of the road data recognition model to obtain the recognition calibration range includes: The data processing resolution of the road data recognition model is obtained. The horizontal pixel value of the data processing resolution is compared with the first preset data to obtain the upper margin. The vertical pixel value of the data processing resolution is subtracted from the second preset value to obtain the lower margin. Set a precision threshold for disease type identification accuracy, and generate a corresponding identification calibration range based on the precision threshold, the upper margin, and the lower margin.

[0040] Specifically, the process of filtering the road defect data to obtain target road defect data, and obtaining road defect inspection results based on the target road defect data, includes: The coordinates of the center coordinate point of the defined identification range are calculated to obtain the center coordinate point value, and corresponding data filtering conditions are formulated based on the center coordinate point value and the defined identification range. The road defect data is filtered according to the data filtering conditions to obtain target road defect data. The severity of the road defect is analyzed according to the road defect specifications of the target road defect data to obtain the severity of the hazard. The road defect inspection results are obtained based on the target road defect data and the severity of the hazard. The data filtering conditions are that the center coordinate point value is greater than the upper edge of the identification range, the center coordinate point value is less than the lower edge of the identification range, and the type identification accuracy is greater than the accuracy threshold of the identification range.

[0041] The process of filtering the road defect data to obtain target road defect data, and then obtaining road defect inspection results based on the target road defect data, further includes: Obtain the inspection results of all roads in the target area, compile the inspection results of all roads into a table, obtain a road disease statistics table, and store the road disease statistics table in the road disease database.

[0042] The present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium stores a road defect inspection program based on visual recognition, and the road defect inspection program based on visual recognition implements the steps of the road defect inspection method based on visual recognition as described above when executed by a processor.

[0043] In summary, this invention provides a road defect inspection method, system, terminal, and storage medium based on visual recognition. The method includes: acquiring road video data of a target area; preprocessing the road video data to obtain target road video data; inputting the target road video data into a trained road data recognition model to obtain road recognition results; and labeling the road recognition results to obtain road defect data; filtering the road defect data to obtain target road defect data; and obtaining road defect inspection results based on the target road defect data. This invention can intelligently identify road defects and intelligently classify road defect types, improving the inspection efficiency and accuracy of road defect inspection results, thereby enabling timely road maintenance.

[0044] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0045] Of course, those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware (such as a processor, controller, etc.). The program can be stored in a computer-readable storage medium, and when executed, it can include the processes described in the above method embodiments. The computer-readable storage medium can be a memory, magnetic disk, optical disk, etc.

[0046] It should be understood that the application of the present invention is not limited to the examples above. Those skilled in the art can make improvements or modifications based on the above description, and all such improvements and modifications should fall within the protection scope of the appended claims.

Claims

1. A road defect inspection method based on visual recognition, characterized in that, The road defect inspection method based on visual recognition includes: Acquire road video data of the target area, and preprocess the road video data to obtain target road video data; The target road video data is input into a trained road data recognition model to obtain road recognition results, and the road recognition results are labeled to obtain road defect data. The road defect data is filtered to obtain target road defect data, and the road defect inspection results are obtained based on the target road defect data.

2. The road defect inspection method based on visual recognition according to claim 1, characterized in that, The process of acquiring road video data of the target area and preprocessing the road video data to obtain target road video data includes, prior to: Acquire a dataset of historical road images and create a data recognition neural network with a preset resolution; The resolution of the historical road image dataset is converted and adjusted according to the preset resolution to obtain the target historical road image dataset. The data recognition neural network is then trained based on the target historical road image dataset to obtain the road data recognition model.

3. The road defect inspection method based on visual recognition according to claim 2, characterized in that, The process of acquiring road video data of the target area and preprocessing the road video data to obtain target road video data specifically includes: The target road in the target area is obtained, road video data of the target road is collected, and the data processing format and data processing resolution of the road data recognition model are obtained. The road video data is converted according to the data processing format to obtain road video data in a preset format. The resolution of the road video data in the preset format is adjusted according to the data processing resolution to obtain the target road video data.

4. The road defect inspection method based on visual recognition according to claim 1, characterized in that, The process of inputting the target road video data into a trained road data recognition model to obtain road recognition results, and then labeling the road recognition results to obtain road defect data, specifically includes: The recognition range is defined according to the data processing resolution of the road data recognition model to obtain the recognition calibration range, and the target road video data is input into the road data recognition model. The road data recognition model performs recognition processing on the target road video data according to the recognition calibration range to obtain road recognition results, wherein the road recognition results include the corner coordinates of the road image, the recognition type, and the type recognition accuracy; The identification type and the identification accuracy are labeled to obtain the road defect type and road defect specification, and road defect data is obtained based on the road defect type and the road defect specification.

5. The road defect inspection method based on visual recognition according to claim 4, characterized in that, The step of defining the recognition range based on the data processing resolution of the road data recognition model to obtain the recognition calibration range specifically includes: The data processing resolution of the road data recognition model is obtained. The horizontal pixel value of the data processing resolution is compared with the first preset data to obtain the upper margin. The vertical pixel value of the data processing resolution is subtracted from the second preset value to obtain the lower margin. Set a precision threshold for disease type identification accuracy, and generate a corresponding identification calibration range based on the precision threshold, the upper margin, and the lower margin.

6. The road defect inspection method based on visual recognition according to claim 4, characterized in that, The process of filtering the road defect data to obtain target road defect data, and then obtaining road defect inspection results based on the target road defect data, specifically includes: The coordinates of the center coordinate point of the defined identification range are calculated to obtain the center coordinate point value, and corresponding data filtering conditions are formulated based on the center coordinate point value and the defined identification range. The road defect data is filtered according to the data filtering conditions to obtain target road defect data. The severity of the road defect is analyzed according to the road defect specifications of the target road defect data to obtain the severity of the hazard. The road defect inspection results are obtained based on the target road defect data and the severity of the hazard. The data filtering conditions are that the center coordinate point value is greater than the upper edge of the identification range, the center coordinate point value is less than the lower edge of the identification range, and the type identification accuracy is greater than the accuracy threshold of the identification range.

7. The road defect inspection method based on visual recognition according to claim 1, characterized in that, The process of filtering the road defect data to obtain target road defect data, and then obtaining road defect inspection results based on the target road defect data, further includes: Obtain the inspection results of all roads in the target area, compile the inspection results of all roads into a table, obtain a road disease statistics table, and store the road disease statistics table in the road disease database.

8. A road defect inspection system based on visual recognition, characterized in that, The vision-based road defect inspection system includes: The data processing module is used to acquire road video data of the target area, preprocess the road video data, and obtain target road video data. The data recognition module is used to input the target road video data into the trained road data recognition model to obtain the road recognition result, and to annotate the road recognition result to obtain road defect data; The road defect inspection module is used to filter the road defect data to obtain target road defect data, and to obtain road defect inspection results based on the target road defect data.

9. A terminal, characterized in that, The terminal includes a memory, a processor, and a program stored in the memory and executable on the processor. When executed by the processor, the program implements the steps of the road defect inspection method based on visual recognition as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, which stores a road defect inspection program based on visual recognition. When the road defect inspection program based on visual recognition is executed by a processor, it implements the steps of the road defect inspection method based on visual recognition as described in any one of claims 1-7.

Images