Large model-based road detection report generation method, apparatus and device, and medium

Abstract

The invention discloses a road detection report generation method and device based on a large model, equipment and a medium, and relates to the technical field of artificial intelligence, and the method comprises the steps: obtaining road detection data based on a pre-trained preset large model, extracting target detection data from the road detection data, and obtaining a road condition diagram based on the target detection data; classifying road disease data in the target detection data by using a preset large model, performing semantic understanding operation on the classified target detection data to obtain a semantic understanding result, and generating an initial road detection report based on the semantic understanding result and the road condition diagram; and optimizing the initial road detection report to obtain a target road detection report according to target feedback information for the initial road detection report obtained from the user side and based on a preset large model. According to the invention, the generation efficiency and quality of the road detection report are improved.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence technology, and particularly to a method, device, equipment and medium for generating road detection reports based on large models. Background Art

[0002] With the continuous growth of highway traffic volume and the increase of road service life, the problem of road diseases has become increasingly prominent, and the demand for road maintenance has also increased. Road detection is the premise and foundation of road maintenance. By regularly detecting the road, the technical condition of the road can be understood in time, road diseases can be discovered, and a scientific basis can be provided for formulating a reasonable maintenance plan.

[0003] Traditional road detection mainly relies on manual work. Detection personnel need to carry detection equipment to the site for data collection and manually analyze the data to generate road detection reports. However, manual detection requires a large amount of time and manpower, the detection speed is slow, and it is difficult to meet the needs of large-scale road detection; when manually analyzing data, it is easily affected by subjective factors, which may lead to incomplete information extraction and omission of important information; and it is difficult to conduct in-depth analysis of manually analyzed data, the analysis conclusion is not accurate enough, and the maintenance plan is not scientific and reasonable enough; at the same time, manually generating reports requires a large amount of time, resulting in too long report generation time and difficult to meet the timeliness requirements.

[0004] With the continuous development of artificial intelligence technology and deep learning, some report generation systems and methods have been proposed. These systems can be used to automatically identify road surface diseases and generate preliminary report content. However, due to insufficient training data volume and limited model generalization ability, it is difficult to adapt to complex and changeable road conditions; at the same time, the system mainly focuses on the identification of road surface diseases, lacks the comprehensive analysis ability of factors such as road structure, traffic load, and climate environment, and it is difficult to give scientific and reasonable maintenance suggestions; and the generated report format is fixed and cannot be customized according to user needs, making it difficult to meet personalized needs. Summary of the Invention

[0005] In view of this, the purpose of this application is to provide a method, device, equipment and medium for generating road detection reports based on large models, which can improve the generation efficiency and quality of road detection reports and provide more accurate data support for road maintenance design. The specific solutions are as follows:

[0006] In the first aspect, this application provides a method for generating a road detection report based on a large model, including:

[0007] Obtaining road detection data based on a preset large model after pre-training, extracting target detection data from the road detection data, and obtaining a road condition map based on the target detection data;

[0008] Classify the road defect data in the target detection data using the preset large model, perform semantic understanding operations on the classified target detection data to obtain a semantic understanding result, and generate an initial road detection report based on the semantic understanding result and the road condition map;

[0009] Optimize the initial road detection report based on the target feedback information obtained from the user terminal for the initial road detection report and using the preset large model to obtain a target road detection report.

[0010] Optionally, the method for generating a road detection report based on a large model further includes:

[0011] Perform initial training on the preset large model based on a preset text data set so that the preset large model learns the language structure and semantics in the preset text data set;

[0012] Train the preset large model after initial training using historical road detection data and historical road detection reports to obtain a pre-trained preset large model.

[0013] Optionally, the target detection data includes road defect data and road environment data; wherein, the road defect data includes road surface defect type data, road surface defect distribution area data, and road surface defect severity data; the road environment data includes road structure data, road traffic load data, and road climate environment data.

[0014] Optionally, extracting target detection data from the road detection data includes:

[0015] Perform image denoising operations and feature enhancement operations on the image data in the road detection data to obtain processed detection data;

[0016] Extract the road defect data and the road environment data from the processed detection data.

[0017] Optionally, optimizing the initial road detection report based on the target feedback information obtained from the user terminal for the initial road detection report and using the preset large model to obtain a target road detection report includes:

[0018] Generate real-time format editing suggestions for the initial road detection report through the preset large model and send the real-time format editing suggestions to the user terminal;

[0019] Obtain the report optimization requirement information for the initial road detection report fed back by the user terminal based on the real-time format editing suggestions;

[0020] Optimize the requirement information based on the report, and optimize the initial road detection report according to the automatic correction function provided by the preset large model to obtain a target road detection report;

[0021] Among them, the report optimization requirement information includes report content optimization requirement information and / or report format optimization requirement information.

[0022] Optionally, the method for generating a road detection report based on a large model further includes:

[0023] Generate the first summary content corresponding to the initial road detection report and / or the second summary content corresponding to the target road detection report through the preset large model;

[0024] Obtain a report viewing request triggered by the user terminal for the initial road detection report and / or the target road detection report, and respond to the report viewing request by using the corresponding first summary content and / or the second summary content.

[0025] Optionally, after optimizing the initial road detection report to obtain a target road detection report, it further includes:

[0026] Construct training data based on the road detection data and the target road detection report;

[0027] Train the preset large model based on the training data to obtain the updated preset large model and the corresponding updated local knowledge base.

[0028] In a second aspect, the present application provides a device for generating a road detection report based on a large model, including:

[0029] A data acquisition module, configured to acquire road detection data based on a preset large model after pre-training, extract target detection data from the road detection data, and obtain a road condition map based on the target detection data;

[0030] A first report generation module, configured to classify road disease data in the target detection data by using the preset large model, perform a semantic understanding operation on the classified target detection data to obtain a semantic understanding result, and generate an initial road detection report based on the semantic understanding result and the road condition map;

[0031] A second report generation module, configured to optimize the initial road detection report according to the target feedback information for the initial road detection report obtained from the user terminal and based on the preset large model to obtain a target road detection report.

[0032] In a third aspect, the present application provides an electronic device, including:

[0033] A memory for storing a computer program;

[0034] A processor for executing the computer program to implement the foregoing method for generating a road detection report based on a large model.

[0035] In a fourth aspect, the present application provides a computer-readable storage medium for storing a computer program, wherein when the computer program is executed by a processor, the foregoing method for generating a road detection report based on a large model is implemented.

[0036] In the present application, road detection data is obtained based on a preset large model after pre-training, target detection data is extracted from the road detection data, and a road condition map is obtained based on the target detection data; the preset large model is used to classify road disease data in the target detection data, a semantic understanding operation is performed on the classified target detection data to obtain a semantic understanding result, and an initial road detection report is generated based on the semantic understanding result and the road condition map; according to target feedback information for the initial road detection report obtained from a user terminal and based on the preset large model, the initial road detection report is optimized to obtain a target road detection report. As can be seen from the above, the present application can automatically process a large amount of road detection data based on a preset large model, improving the detection speed, shortening the report generation time, improving work efficiency, and meeting the needs of large-scale road detection; using the preset large model to perform a semantic understanding operation on target detection data to obtain a semantic understanding result, and generating a road detection report based on the semantic understanding result. Through the in-depth analysis ability of the large model, key information in the road detection data can be extracted from multiple dimensions, ensuring the comprehensiveness of the generated report content, effectively avoiding omissions and biases that may occur in manual analysis, making the road detection report more detailed and accurate. At the same time, the large model gives scientific and reasonable maintenance suggestions by analyzing the data, improving the pertinence and effectiveness of road maintenance; optimizing the road detection report based on the user's feedback information and the large model to obtain the final target road detection report, meeting the needs of different users and enhancing the practicality and scope of application of the report. Description of the Drawings

[0037] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only the embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained according to the provided drawings without creative efforts.

[0038] Figure 1 It is a flowchart of a method for generating a road detection report based on a large model disclosed in the present application;

[0039] Figure 2 Schematic diagram of a specific method for generating a road detection report based on a large model disclosed in this application;

[0040] Figure 3 Schematic diagram of the structure of a device for generating a road detection report based on a large model disclosed in this application;

[0041] Figure 4 Schematic diagram of the structure of an electronic device disclosed in this application. Specific embodiments

[0042] Next, the technical solutions in the embodiments of this application will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of this application. Obviously, the described embodiments are only a part of the embodiments of this application, rather than all the embodiments. Based on the embodiments in this application, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of this application.

[0043] With the continuous growth of highway traffic volume and the increase in the service life of roads, the problem of road diseases has become increasingly prominent, and the demand for road maintenance has also increased. Traditional road detection reports mainly rely on manual generation, which is not only inefficient but also prone to information omission and analysis deviation, making it difficult to meet the needs of rapid response and scientific decision-making. With the continuous development of artificial intelligence technology and deep learning, some report generation systems and methods have been proposed. However, the system lacks the comprehensive analysis ability of factors such as road structure and traffic load, making it difficult to give scientific and reasonable maintenance suggestions; and the generated report format is fixed, making it difficult to meet personalized needs. Therefore, this application provides a method for generating a road detection report based on a large model, which can improve the generation efficiency and quality of road detection reports and provide more accurate data support for road maintenance design.

[0044] See Figure 1 As shown, the embodiments of this application disclose a method for generating a road detection report based on a large model, including:

[0045] Step S11: Obtain road detection data based on a preset large model after pre-training, extract target detection data from the road detection data, and obtain a road condition map based on the target detection data.

[0046] In this embodiment, first, road detection data is obtained based on a preset large model after pre-training. Among them, first, the preset large model is initially trained using a preset text data set so that the preset large model learns the language structure and semantics in the preset text data set; then, the initially trained preset large model is trained using historical road detection data and historical road detection reports to obtain the preset large model after pre-training.

[0047] It is understandable that a large-scale text dataset is used to initially train a preset large model so that the preset large model can learn the basic structure and semantics of the language, ensuring that the preset large model has strong language understanding ability and laying a foundation for subsequent fine-tuning. On the basis of the initial training, the preset large model is fine-tuned for specific tasks so that the preset large model can adapt to the specific requirements of road detection report generation. During the fine-tuning process, the preset large model is further trained based on a dataset containing road detection data and historical road detection reports, thereby optimizing the preset large model's understanding of the road detection field and its ability to generate road detection reports. Through fine-tuning, the preset large model can more accurately identify road diseases and extract valuable information for road detection report generation.

[0048] Then, based on the preset large model after pre-training, image denoising operations and feature enhancement operations are performed on the image data in the road detection data to obtain processed detection data, and target detection data such as road disease data and road environment data are extracted from the processed detection data. Among them, the target detection data includes, but is not limited to, road disease data and road environment data; the above-mentioned road disease data includes, but is not limited to, road surface disease type data, road surface disease distribution area data, and road surface disease severity data; the above-mentioned road environment data includes, but is not limited to, road structure data, road traffic load data, and road climate environment data.

[0049] Among them, the key features in the road detection data can be identified and extracted from the processed detection data based on the preset large model. For example, disease features such as cracks and potholes can be identified from images, or features related to road use and environmental impacts can be extracted from sensor data, and then target detection data such as road disease data and road environment data can be extracted.

[0050] It is understandable that the above-mentioned road detection data can include road visual images captured by cameras, which visually show the surface conditions of the road, such as disease conditions like cracks, potholes, and ruts on the road surface; it can also include the sub-surface structure information of the road surface detected by radar equipment, such as the condition of the roadbed, underground cavities, or loose areas, or information such as the speed, distance, and driving trajectory of vehicles on the road monitored; during the data collection process, images are easily interfered by various factors and generate noise. Therefore, the obtained road detection data can be pre-processed based on the preset large model, such as using the preset large model for image denoising, feature enhancement, etc. to restore the true information of the image, or corresponding pre-processing operations can be performed on other types of data in the road detection data, such as radar data and sensor data, to improve the accuracy of subsequent analysis.

[0051] After obtaining the target detection data, further, a road condition map is obtained based on the target detection data. Specifically, by leveraging the powerful learning ability of the preset large model, the target detection data is learned to discover the hidden correlations between different data sources. For example, there may be a correlation between certain specific radar signal features and a certain disease on the road surface in the visual image. Further, the preset large model uses the learned correlations to integrate and fuse the data from different sources, combining the advantages of each data source to form a comprehensive and accurate view of the road condition. This road condition view is not limited to the information provided by a single data source but combines the characteristics of multiple data sources, enabling a deeper and more complete reflection of the actual road condition and providing stronger support for road maintenance, management, and decision-making.

[0052] Step S12: Use the preset large model to classify the road disease data in the target detection data, perform semantic understanding operations on the classified target detection data to obtain a semantic understanding result, and generate an initial road detection report based on the semantic understanding result and the road condition map.

[0053] In this embodiment, first, the preset large model is used to classify the road disease data in the target detection data. In actual road detection work, manual identification of diseases is not only inefficient but also may be subjective and error-prone. By leveraging the powerful computing and learning ability of the preset large model, the road diseases can be quickly and accurately judged and classified. It should be noted that as new road disease data is continuously collected, the preset large model can use this new data for further learning and training to adjust its own parameters and algorithm structure, thereby improving the accuracy and robustness of disease detection.

[0054] After classifying the road disease data, semantic understanding operations can be performed on the classified target detection data to deeply explore the meanings and relationships behind the data. For example, the preset large model can analyze the correlations between different diseases to determine whether the occurrence of a certain disease will trigger other diseases; or analyze the mutual influence between road environment data and road diseases. Through semantic understanding, a more comprehensive and in-depth understanding result of the road condition can be obtained.

[0055] Further, an initial road detection report is generated based on the semantic understanding result and the road condition map generated in step S11. It can be understood that the road condition map intuitively displays various information about the road, such as the location of diseases, the structure of the road, etc.; the semantic understanding result provides a deeper analysis and interpretation of this information. Based on the semantic understanding result and the road condition map, the overall condition of the road can be obtained, and then the initial road detection report can be obtained.

[0056] Step S13: Optimize the initial road detection report based on the target feedback information obtained from the client side and based on the preset large model to obtain a target road detection report.

[0057] In this embodiment, optimizing the initial road detection report based on the target feedback information obtained from the client side and based on the preset large model to obtain a target road detection report may include: First, generate real-time format editing suggestions for the initial road detection report through the preset large model and send the real-time format editing suggestions to the client side; then obtain the report optimization requirement information for the initial road detection report fed back by the client side based on the real-time format editing suggestions; further, optimize the initial road detection report based on the report optimization requirement information and according to the automatic correction function provided by the preset large model to obtain a target road detection report; where the report optimization requirement information includes report content optimization requirement information and / or report format optimization requirement information.

[0058] It can be understood that if the preset large model finds that the layout of some parts of the report does not conform to industry standards, or the data presentation method is not clear and intuitive enough, it generates corresponding format adjustment suggestions, such as adjusting paragraph spacing, unifying title formats, optimizing table styles, etc., and then sends the format adjustment suggestions to the client side, so that the user can understand the possible problems and improvement directions of the report in terms of format. After receiving the real-time format editing suggestions sent by the preset large model, the client side further examines and evaluates the initial road detection report, and combines its own needs and actual situations to feedback the report optimization requirement information for the initial road detection report. The report optimization requirement information may include report content optimization requirement information, such as believing that the analysis of some diseases is not deep enough and the explanation of some data is not clear enough, and more detailed information needs to be supplemented; it may also include report format optimization requirement information, that is, the user has further requirements for the format suggestions put forward by the preset large model, or has its own unique format preferences. After obtaining the report optimization requirement information fed back by the client side, optimize the initial road detection report based on the automatic correction function provided by the preset large model. If it is a report content optimization requirement, the preset large model can use its knowledge reserve and reasoning ability to fill in information or deepen the analysis of the parts that need to be supplemented or improved in the report. For example, for the problem of insufficient in-depth analysis of diseases, the preset large model can search for relevant materials and cases to supplement more information about the causes, development trends, and possible impacts of diseases; if it is a report format optimization requirement, the preset large model will adjust the format of the report according to the user's requirements, such as changing the chart style and adjusting the text layout. After the above series of optimization operations, a target road detection report that meets the user's needs is finally obtained.

[0059] It should be noted that the model can be trained based on road detection data and the corresponding target road detection report to update the knowledge base of the model. Specifically, first, training data is constructed based on the road detection data and the target road detection report; then, the pre-set large model is trained based on the training data to obtain the updated pre-set large model and the corresponding updated local knowledge base. This process enables the large model to more accurately analyze the causes of road diseases and provide scientific and reasonable suggestions for road maintenance in combination with the latest research results and technological trends.

[0060] In this embodiment, to increase the readability of the report, the first summary content corresponding to the initial road detection report and / or the second summary content corresponding to the target road detection report can be generated by the pre-set large model; and when a report viewing request for the initial road detection report and / or the target road detection report triggered by the user side is obtained, the corresponding first summary content and / or second summary content is used to respond to the report viewing request.

[0061] The method of generating a summary by the pre-set large model and responding according to the user request not only improves the readability of the road detection report but also optimizes the user experience of viewing report information.

[0062] As can be seen from the above, this embodiment can automatically process a large amount of road detection data based on the pre-set large model, improve the detection speed, shorten the time for report generation, improve work efficiency, and meet the needs of large-scale road detection; use the pre-set large model to perform semantic understanding operations on the target detection data to obtain a more comprehensive and in-depth understanding result of the road condition, and generate a road detection report based on the semantic understanding result. Through the deep analysis ability of the large model, key information in the road detection data can be extracted from multiple dimensions to ensure the comprehensiveness of the generated report content, effectively avoiding omissions and biases that may occur in manual analysis, making the road detection report more detailed and accurate. At the same time, the large model gives scientific and reasonable maintenance suggestions by analyzing the data, improving the pertinence and effectiveness of road maintenance; optimizing the road detection report based on the report optimization requirement information fed back by the user side and the large model to obtain the target road detection report that meets the user's needs, enhancing the practicability and scope of application of the report.

[0063] The following takes Figure 2 as an example to illustrate the technical solution in this application.

[0064] First, the pre-set large model is initially trained based on the pre-set text dataset so that the pre-set large model learns the language structure and semantics in the pre-set text dataset; then, the pre-set large model after initial training is fine-tuned for specific tasks using historical road detection data and historical road detection reports to obtain the pre-set large model after pre-training.

[0065] Then, road detection data processing is carried out. Specifically, first, based on a preset large model, data preprocessing is performed on the obtained road detection data. For example, operations such as image denoising and feature enhancement are carried out using the preset large model to restore the true information of the image. Corresponding preprocessing operations can also be performed on other types of data in the road detection data, such as radar data and sensor data, to obtain the processed detection data. Then, based on the preset large model, key features in the road detection data are identified and extracted from the processed detection data, and then target detection data such as road disease data and road environment data are extracted. Further, using the powerful learning ability of the preset large model, the target detection data is learned to discover the hidden correlations between different data sources, so as to combine the advantages of each data source to form a comprehensive and accurate view of the road condition. Finally, the preset large model is used to classify the road disease data in the target detection data.

[0066] Further, a road detection report is generated. Specifically, based on the preset large model, semantic understanding operations are performed on the classified target detection data to obtain a semantic understanding result. Based on the semantic understanding result and the road condition map, the overall condition of the road is obtained, and then an initial road detection report is generated, and a corresponding first summary content is generated for the initial road detection report. Then, real-time format editing suggestions for the initial road detection report are generated through the preset large model, and the real-time format editing suggestions are sent to the user terminal, and the report content optimization requirement information and report format optimization requirement information for the initial road detection report feedback by the user terminal based on the real-time format editing suggestions are obtained. Further, based on the report content optimization requirement information and the report format optimization requirement information, the initial road detection report is optimized according to the automatic correction function provided by the preset large model to obtain the target road detection report, and a corresponding second summary content is generated for the target road detection report.

[0067] Finally, training data can be constructed based on the road detection data and the corresponding target road detection report, and then the training data is used to train the preset large model to update the local knowledge base of the model.

[0068] As can be seen from the above, in this embodiment, the preset large model is used to quickly process the road detection data, improve the detection speed, meet the needs of large-scale road detection, and realize the automatic recognition of road diseases and the intelligent generation of reports. At the same time, through the comprehensive analysis of multiple factors such as road structure, traffic load, and climate environment, scientific and reasonable maintenance suggestions are given to improve the accuracy of the maintenance plan. In addition, the report format and content can be customized according to user needs to meet personalized needs, shorten the report generation time, improve the timeliness of the report, and effectively improve the efficiency and quality of road maintenance work.

[0069] See Figure 3As shown in the figure, an embodiment of the present application also discloses a road detection report generation device based on a large model, including:

[0070] A data acquisition module 11, configured to obtain road detection data based on a preset large model after pre-training, extract target detection data from the road detection data, and obtain a road condition map based on the target detection data;

[0071] A first report generation module 12, configured to classify the road disease data in the target detection data by using the preset large model, perform a semantic understanding operation on the classified target detection data to obtain a semantic understanding result, and generate an initial road detection report based on the semantic understanding result and the road condition map;

[0072] A second report generation module 13, configured to optimize the initial road detection report based on the target feedback information obtained from the user side for the initial road detection report and based on the preset large model to obtain a target road detection report.

[0073] As can be seen from the above, the present application can automatically process a large amount of road detection data based on a preset large model, improve the detection speed, shorten the report generation time, improve work efficiency, and meet the needs of large-scale road detection; use the preset large model to perform a semantic understanding operation on the target detection data to obtain a semantic understanding result, and generate a road detection report based on the semantic understanding result. Through the in-depth analysis ability of the large model, key information in the road detection data can be extracted from multiple dimensions to ensure the comprehensiveness of the generated report content, effectively avoiding omissions and biases that may occur in manual analysis, making the road detection report more detailed and accurate. At the same time, the large model gives scientific and reasonable maintenance suggestions by analyzing the data, improving the pertinence and effectiveness of road maintenance; optimizing the road detection report based on the user's feedback information and the large model to obtain the final target road detection report, meeting the needs of different users, and enhancing the practicability and application scope of the report.

[0074] In some specific embodiments, the road detection report generation device based on a large model further includes:

[0075] A first model training unit, configured to perform initial training on a preset large model based on a preset text data set, so that the preset large model learns the language structure and semantics in the preset text data set;

[0076] A second model training unit, configured to train the preset large model after initial training by using historical road detection data and historical road detection reports to obtain a preset large model after pre-training.

[0077] In some specific embodiments, the target detection data includes road disease data and road environment data; wherein, the road disease data includes road surface disease type data, road surface disease distribution area data, and road surface disease severity data; and the road environment data includes road structure data, road traffic load data, and road climate environment data.

[0078] In some specific embodiments, the data acquisition module 11 includes:

[0079] A data processing unit, configured to perform image denoising operations and feature enhancement operations on the image data in the road detection data to obtain processed detection data;

[0080] A data extraction unit, configured to extract the road disease data and the road environment data from the processed detection data.

[0081] In some specific embodiments, the second report generation module 13 includes:

[0082] A suggestion sending unit, configured to generate real-time format editing suggestions for the initial road detection report through the preset large model and send the real-time format editing suggestions to the user terminal;

[0083] An information acquisition unit, configured to acquire report optimization requirement information for the initial road detection report fed back by the user terminal based on the real-time format editing suggestions;

[0084] A report optimization unit, configured to optimize the initial road detection report based on the report optimization requirement information and according to the automatic correction function provided by the preset large model to obtain a target road detection report;

[0085] Wherein, the report optimization requirement information includes report content optimization requirement information and / or report format optimization requirement information.

[0086] In some specific embodiments, the road detection report generation device based on a large model further includes:

[0087] An abstract generation unit, configured to generate a first abstract content corresponding to the initial road detection report and / or a second abstract content corresponding to the target road detection report through the preset large model;

[0088] A request response unit, configured to acquire a report viewing request triggered by the user terminal for the initial road detection report and / or the target road detection report, and respond to the report viewing request by using the corresponding first abstract content and / or second abstract content.

[0089] In some specific embodiments, the second report generation module 13 further includes:

[0090] A data construction unit, configured to construct training data based on the road detection data and the target road detection report;

[0091] A model training unit, configured to train the preset large model based on the training data to obtain the updated preset large model and the corresponding updated local knowledge base.

[0092] Furthermore, an embodiment of the present application also discloses an electronic device. Figure 4 It is a structural diagram of an electronic device 20 shown according to an exemplary embodiment. The content in the figure should not be considered as any limitation to the scope of use of the present application.

[0093] Figure 4 This is a schematic structural diagram of an electronic device 20 provided by an embodiment of the present application. The electronic device 20 may specifically include: at least one processor 21, at least one memory 22, a power supply 23, a communication interface 24, an input / output interface 25, and a communication bus 26. Among them, the memory 22 is used to store a computer program, and the computer program is loaded and executed by the processor 21 to implement the relevant steps in the method for generating a road detection report based on a large model disclosed in any of the foregoing embodiments. In addition, the electronic device 20 in this embodiment may specifically be an electronic computer.

[0094] In this embodiment, the power supply 23 is used to provide operating voltage for each hardware device on the electronic device 20; the communication interface 24 can create a data transmission channel between the electronic device 20 and external devices, and the communication protocol it follows is any communication protocol applicable to the technical solution of the present application, and specific limitations are not imposed here; the input / output interface 25 is used to obtain external input data or output data to the outside, and its specific interface type can be selected according to specific application needs, and specific limitations are not imposed here.

[0095] In addition, as a carrier for resource storage, the memory 22 may be a read-only memory, a random access memory, a disk, or an optical disc, etc. The resources stored thereon may include an operating system 221, a computer program 222, etc., and the storage method may be short-term storage or permanent storage.

[0096] Among them, the operating system 221 is used to manage and control each hardware device and computer program 222 on the electronic device 20, and it can be Windows Server, Netware, Unix, Linux, etc. In addition to the computer program that can be used to complete the method for generating a road detection report based on a large model executed by the electronic device 20 disclosed in any of the foregoing embodiments, the computer program 222 can further include computer programs that can be used to complete other specific tasks.

[0097] Furthermore, the present application also discloses a computer-readable storage medium for storing a computer program; wherein, when the computer program is executed by a processor, it implements the method for generating a road detection report based on a large model disclosed above. For the specific steps of this method, reference can be made to the corresponding content disclosed in the foregoing embodiments, and details will not be elaborated herein.

[0098] In this specification, the various embodiments are described in a progressive manner. Each embodiment focuses on the differences from other embodiments. For the same or similar parts among the various embodiments, reference can be made to each other. For the devices disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple, and reference can be made to the description of the method part for related parts.

[0099] Those skilled in the art can further realize that the units and algorithm steps of the examples described in conjunction with the embodiments disclosed herein can be implemented by electronic hardware, computer software, or a combination of the two. To clearly illustrate the interchangeability of hardware and software, the composition and steps of the examples have been generally described according to functions in the above description. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present application.

[0100] The steps of the method or algorithm described in conjunction with the embodiments disclosed herein can be directly implemented by hardware, a software module executed by a processor, or a combination of the two. The software module can be placed in a random access memory (RAM), memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, register, hard disk, removable disk, CD-ROM, or any other form of storage medium well-known in the technical field.

[0101] Finally, it should also be noted that in this text, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements not only includes those elements, but also includes other elements not expressly listed, or further includes elements inherent to such process, method, article or device. Without further limitation, an element defined by the statement "comprising an..." does not exclude the presence of additional identical elements in the process, method, article or device comprising the said element.

[0102] The technical solutions provided in this application have been introduced in detail above. Specific examples are used in this text to elaborate on the principle and implementation manner of this application. The description of the above embodiments is only used to help understand the method and its core idea of this application; at the same time, for those of ordinary skill in the art, according to the idea of this application, there will be changes in the specific implementation manner and application scope. In summary, the content of this specification should not be construed as a limitation to this application.

Claims

1. A method for generating a road detection report based on a large model, characterized in that, Including: Obtaining road detection data based on a pre-trained preset large model, extracting target detection data from the road detection data, and obtaining a road condition map based on the target detection data; Using the preset large model to classify road disease data in the target detection data, performing a semantic understanding operation on the classified target detection data to obtain a semantic understanding result, and generating an initial road detection report based on the semantic understanding result and the road condition map; Optimizing the initial road detection report based on the target feedback information obtained from the user side for the initial road detection report and based on the preset large model to obtain a target road detection report.

2. The method for generating a road detection report based on a large model according to claim 1, wherein Also including: Initial training of the preset large model based on a preset text dataset so that the preset large model learns the language structure and semantics in the preset text dataset; Training the preset large model after initial training using historical road detection data and historical road detection reports to obtain a pre-trained preset large model.

3. The method for generating a road detection report based on a large model according to claim 1, wherein The target detection data includes road disease data and road environment data; among them, the road disease data includes road surface disease type data, road surface disease distribution area data, and road surface disease severity data; the road environment data includes road structure data, road traffic load data, and road climate environment data.

4. The method for generating a road detection report based on a large model according to claim 3, wherein, The extracting target detection data from the road detection data includes: Performing an image denoising operation and a feature enhancement operation on the image data in the road detection data to obtain processed detection data; Extracting the road disease data and the road environment data from the processed detection data.

5. The method for generating a road detection report based on a large model according to any one of claims 1 to 4, characterized in that, The optimizing the initial road detection report based on the target feedback information obtained from the user side for the initial road detection report and based on the preset large model to obtain a target road detection report includes: Generating real-time format editing suggestions for the initial road detection report through the preset large model and sending the real-time format editing suggestions to the user side; Obtaining report optimization requirement information for the initial road detection report feedback by the user side based on the real-time format editing suggestions; Optimizing the initial road detection report based on the report optimization requirement information and according to the automatic correction function provided by the preset large model to obtain a target road detection report; Among them, the report optimization requirement information includes report content optimization requirement information and / or report format optimization requirement information.

6. The method for generating a road detection report based on a large model according to claim 5, wherein Also including: Generating a first summary content corresponding to the initial road detection report and / or a second summary content corresponding to the target road detection report through the preset large model; Obtaining a report viewing request triggered by the user side for the initial road detection report and / or the target road detection report, and responding to the report viewing request using the corresponding first summary content and / or second summary content.

7. The method for generating a road detection report based on a large model according to any one of claims 1 to 4, characterized in that After optimizing the initial road detection report to obtain a target road detection report, it also includes: Constructing training data based on the road detection data and the target road detection report; Training the preset large model based on the training data to obtain the updated preset large model and the corresponding updated local knowledge base.

8. An apparatus for generating a road detection report based on a large model, characterized in that, Including: A data acquisition module, configured to obtain road detection data based on a preset large model after pre-training, extract target detection data from the road detection data, and obtain a road condition map based on the target detection data; A first report generation module, configured to classify road disease data in the target detection data by using the preset large model, perform a semantic understanding operation on the classified target detection data to obtain a semantic understanding result, and generate an initial road detection report based on the semantic understanding result and the road condition map; A second report generation module, configured to optimize the initial road detection report based on the preset large model according to the target feedback information for the initial road detection report obtained from the user side to obtain a target road detection report.

9. An electronic device, characterized in that, Including: A memory, configured to store a computer program; A processor, configured to execute the computer program to implement the large model-based road detection report generation method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, For storing a computer program, the computer program, when executed by a processor, implements the large model-based road detection report generation method according to any one of claims 1 to 7.

Images