Municipal engineering pipe network multi-attribute information collecting and reporting method based on large model

Abstract

The invention belongs to the technical field of municipal engineering pipe network file management, and particularly relates to a municipal engineering pipe network multi-attribute information collecting and reporting method based on a large model. A collection APP is used, a multi-mode driving construction pipeline multi-attribute recognition large model is built in the collection APP, and the method comprises the steps that the collection APP is opened and logged in; setting a reference object; standardized photo shooting; constructing a multi-modal data set; automatically identifying multiple attributes of the pipe network; and constructing an archived data set. Compared with the prior art, the method has the beneficial effects that the collection efficiency is improved; the data accuracy is improved; a multi-attribute data record is constructed, the format is uniform, and interaction is convenient; a data encryption storage mode in which item binding, directional uploading multi-level permission control, user password encryption and photo URL are unique and cannot be tampered is adopted, and data leakage or misoperation is prevented; newly added reference object types are supported, identification attributes are expanded, and collection requirements of different municipal pipe networks are met.

Description

Technical Field

[0001] This invention belongs to the field of municipal engineering pipeline network archive management technology, specifically involving a method for collecting and reporting multi-attribute information of municipal engineering pipeline networks based on a large model. Background Technology

[0002] Currently, the collection of municipal engineering pipeline network information mainly relies on manual recording (such as paper ledgers and manual measurement) + subsequent digital entry, which has the following core pain points: 1. Low data collection efficiency: Manually measuring pipe dimensions and recording location information is time-consuming, and photos and data need to be sorted out separately, resulting in a fragmented process; 2. Poor data accuracy: Pipe dimensions rely on manual estimation or measurement with simple tools, resulting in large errors; there is no uniform standard for taking photos, which can easily lead to the loss of key scene information and affect subsequent record tracing. 3. Cumbersome data interaction: Collected data needs to be manually converted into a format before being imported into the file system. Data loss or mismatch is prone to occur during secondary processing. 4. Lack of access control: Construction personnel can operate across projects, posing a risk of data leakage or accidental modification.

[0003] While existing technologies include mobile data collection apps, most do not integrate with large models for automated recognition and lack direct integration with archiving systems, thus failing to meet the municipal engineering requirements for "efficient, accurate, and seamless archiving" of pipeline network information. Summary of the Invention

[0004] To address the problems existing in the prior art, this invention provides a method for collecting and reporting multi-attribute information of municipal engineering pipeline networks based on large models. It is a whole-process method that integrates "mobile terminal collection - cloud-based large model recognition - direct connection to the archive system" to solve the above problems.

[0005] The technical solution adopted by this invention to solve its technical problem is as follows: A method for collecting and reporting multi-attribute information of municipal engineering pipeline networks based on a large model, using a collection APP, the collection APP having a built-in multi-modal driven construction pipeline multi-attribute identification large model, including the following steps: S1. Open and log in to the data collection app; S2. Reference object setting: Place the reference object according to the voice prompts and match the reference object size parameters; S3. Standardized photo shooting: Take photos in a fixed order according to voice prompts and upload them; S4. Multimodal dataset construction: Construct a multimodal dataset in the format of "reference object size + compliant photo URL + user ID + project ID", transmit it to the backend, and the backend returns "recognition task ID"; S5. Automatic identification of multiple attributes of pipeline network, multimodal driven construction pipeline multi-attribute identification large model is based on the size of reference object and photo features, calculates and returns pipeline network attribute information, with no missing information, and uploads it; S6. Constructing the archived dataset: The APP automatically collects location data, time data, personnel data, and attachment data. The data is encapsulated into JSON format according to the requirements of the archive system. The encapsulated data includes three sub-objects: basic collection information, pipeline attribute information, and attachment information.

[0006] Preferably, it also includes the deployment and configuration of the data collection APP, the deployment and interface configuration of the cloud recognition model, and the adaptation of the archive system interface.

[0007] Preferably, the deployment and configuration of the data collection app includes: GPS permission management, intelligent photo processing, and targeted upload control.

[0008] Preferably, GPS permission management: When the APP is launched for the first time, a pop-up window requests location permission to ensure that the latitude and longitude accuracy of the collection point is ≤10 meters, providing a basis for pipeline location tracing.

[0009] Preferably, the photo is intelligently processed: it automatically adapts to the phone's camera resolution, compresses the photo to 1-2MB after shooting, and limits the format to JPG / PNG.

[0010] Preferably, targeted upload control: the APP only transmits data to the preset FLASK backend dedicated interface and prohibits sending to third-party addresses.

[0011] Preferably, the reference object location requirements are as follows: the reference object must be placed on the same horizontal side as the laid pipeline, at a distance of ≤1 meter from the pipeline, to ensure that the pipeline and the reference object can be captured simultaneously when the photo is taken.

[0012] Preferably, the standardized photographs include: panoramic photos, side photos, and close-up photos.

[0013] Preferably, the pipeline attribute information includes: pipeline width, pipeline height above ground, pipeline intersection situation and number of intersections.

[0014] Preferably, the data collection app obtains the longitude and latitude of the current collection point through the GPS module to obtain location data; the data collection app obtains the current system time as time data; the data collection app obtains the currently logged-in operator's account information as personnel data; and the data collection app uses the URL of the compliant photo as attachment data.

[0015] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. Improved data collection efficiency: The standardized data collection process, which includes voice-guided shooting, automatic verification, and direct connection to the archive system, eliminates the need for manual measurement, recording, and format conversion, reducing the data collection time per point from the traditional 30 minutes to within 5 minutes.

[0016] 2. Improved data accuracy: A large model for multi-attribute identification of construction pipelines is adopted, which is based on known size reference objects. The pipeline size identification error is ≤0.5cm, which is far lower than the 5-10cm error of manual estimation. It also avoids the problem of missing key information in photos and facilitates the traceability of archives in the later stage.

[0017] 3. Construct multi-attribute data records with a unified format for easy interaction.

[0018] 4. Security Guaranteed: Employs multi-level access control with project binding and targeted uploads, as well as encrypted data storage methods such as encrypted user passwords and unique, tamper-proof photo URLs to prevent data leakage or accidental operation.

[0019] 5. High scalability: Supports adding new reference object types and expanding recognition attributes to adapt to the data collection needs of different municipal pipe networks. Attached Figure Description

[0020] Figure 1 This is a schematic diagram of the process of the present invention.

[0021] Figure 2 for Figure 1 A flowchart illustrating the process of system configuration in the early and middle stages.

[0022] Figure 3 for Figure 1 A schematic diagram of the process in the large model for multi-attribute identification of multi-modal driven construction pipelines. Detailed Implementation

[0023] To facilitate understanding of the present invention, it will be described in more detail below with reference to the accompanying drawings and specific embodiments. However, the present invention can be implemented in many different forms and is not limited to the embodiments described in this specification. Rather, these embodiments are provided to provide a more thorough and complete understanding of the disclosure of the present invention.

[0024] This method for collecting and reporting multi-attribute information of municipal engineering pipeline networks based on a large model achieves efficient collection and reporting of multi-attribute information through four major stages: "preliminary system and terminal configuration - standardized on-site collection - automated cloud processing - seamless archiving." It relies on a three-layer architecture: terminal layer (collection APP), cloud service layer (FLASK backend + recognition model), and data layer (temporary database + archiving system). Figure 1-3 To understand, the specific steps are as follows: Preliminary preparations: System and terminal configuration 1. Data collection APP deployment and configuration Hardware compatibility: Install the "Construction Pipeline Information Collection APP" (hereinafter referred to as the Collection APP) on the smartphones of construction workers that support Android 10.0 and above or iOS 14.0 and above. Core functions are built-in: GPS Permission Management: When the APP is launched for the first time, a pop-up window requests location permission to ensure that the latitude and longitude accuracy of the collection point is ≤10 meters, providing a basis for pipeline location tracing. Intelligent photo processing: Automatically adapts to the phone's camera resolution (supports devices with ≥13 megapixels), compresses photos to 1-2MB after shooting (balancing clarity and transmission efficiency), and limits the format to JPG / PNG; Targeted upload control: The app only transmits data to the pre-defined FLASK backend dedicated interface, prohibiting sending to third-party addresses to ensure data security.

[0025] 2. Cloud-based recognition model deployment and interface configuration Model Deployment: Deploy the trained "Multimodal Driven Construction Pipeline Multi-Attribute Recognition Large Model" on a Linux server (CentOS 7.6+, 2 cores, 4GB or more of memory); API Development: Two HTTP APIs are exposed via the FLASK backend, supporting JSON data exchange. Photo verification interface: Receives 4 photos uploaded by the APP, verifies whether they are clear and unobstructed, and whether they meet the recognition scenario (e.g., side photos need to include reference objects), with a response time of ≤3 seconds; Multi-attribute recognition interface: Receives a multimodal dataset (reference object size + compliant photo), and returns pipe width, height above ground, pipe intersection status and number of intersections, with 100% field integrity.

[0026] 3. Preparation of standard reference objects and parameter presets Reference object selection: Use spherical reference objects with known precise dimensions, including basketballs (diameter 24.6cm) and soccer balls (diameter 21.6cm), to ensure that the size error is ≤0.1cm; Parameter input: The size data of the two reference objects are preset into the temporary MySQL database in the FLASK backend, and the APP provides corresponding selection options (drop-down menu) to avoid size errors caused by manual input.

[0027] 4. File system interface adaptation Field alignment: The FLASK backend development has a dedicated data transmission interface that strictly matches the fields in the construction unit's "Construction Pipeline Location Information Database", including: GPS longitude (floating-point number), GPS latitude (floating-point number), pipe width (floating-point number), pipe height above ground (floating-point number), pipe intersection status (boolean value), number of intersections (integer), photo URL (string), collection time (timestamp), and operator account information (string). Seamless writing: The interface supports direct writing of data to the archive system without secondary format conversion. After receiving the data, the archive system automatically assigns a unique file number (such as "SG-GW-2025-XXX").

[0028] On-site implementation: Standardized on-site data collection 1. User login and permission verification Login method: Construction workers open the APP and choose one of the two login modes: Username / Password Login: Enter the preset username and password. The FLASK backend queries the "User Information" table (stored in a temporary MySQL database) to verify the validity of the account and its associated "Project ID". Enterprise WeChat Authorized Login: The APP calls the Enterprise WeChat SDK to obtain the user's unique identifier (Enterprise WeChat ID), and the FLASK backend verifies the identity through the Enterprise WeChat open interface and associates the corresponding project permissions; Access control: Users are only allowed to operate the data collection function of their own project. They cannot view / edit data of other projects. When verification fails, the APP will display a pop-up message saying "No permission to operate this project".

[0029] 2. Setting up reference points Operation guidance: The APP provides voice prompts such as "Please place the reference object near the pipeline to ensure that it is fully visible (unobstructed and not tilted)"; Location requirements: The reference object must be placed on the same horizontal side as the laid pipeline, at a distance of ≤1 meter from the pipeline, to ensure that the pipeline and the reference object can be captured simultaneously when taking the photo.

[0030] 3. Standardized photo shooting Shooting guidance: The app provides voice prompts in a fixed order to take 4 photos. If the previous photo is not taken, the app will lock the function to take the next photo. First photo (panoramic view): The lens should cover the pipeline and the surrounding ground environment (such as construction signs and road references), and the location of the pipeline on site should be clearly shown; Photos 2-3 (side views): Take photos from two vertical sides of the pipe trench, ensuring that the reference object and the cross-section of the pipe trench are fully included, and that the reference object is unobstructed and the pipe outline is clear. Image 4 (close-up): The lens focuses on the pipe body, capturing details such as the outer wall and interfaces of the pipe, with a resolution of ≥1080P, for attribute recognition; Automatic processing: After shooting, the APP automatically compresses the photo to 1-2MB and marks the photo type ("panorama", "side view 1", "side view 2", "close-up").

[0031] 4. Photo Upload and Smart Verification Data Upload: When the construction worker clicks "Upload Verification", the APP will transmit 4 photos with different tags to the FLASK backend through the "Photo Upload Verification Interface" (POST request, carrying token, projectID, photos file, and photoTypes parameters); Model Validation: The FLASK backend forwards the photo to the recognition model, and the model is validated in the following dimensions: Clarity: The photo is clear and free of motion blur; Completeness: Panoramic photos include the surrounding environment of the pipeline, side photos include reference objects, and close-up photos include pipeline details; Compliance: Photos must be in JPG / PNG format and each photo must be ≤5MB in size; Feedback results: If the verification passes, the FLASK backend returns "checkResult: Success"; if the verification fails, a specific error message is returned (such as "The side view photo 1 does not contain a reference object, please take a new photo"), and the APP will provide a pop-up message with voice and text.

[0032] 5. Construction of multimodal datasets Reference point selection: The APP pops up a reference point selection interface ("basketball" or "soccer ball"), and the construction personnel select a reference point that is consistent with the site. Size retrieval: The APP retrieves the standard size of the selected reference object from the FLASK backend via an interface (to avoid local storage tampering). Dataset encapsulation: Construct a multimodal dataset in the format of "reference object size (e.g., 24.6cm) + URLs of 4 compliant photos + user ID + project ID", and transmit it to the FLASK backend through the "multimodal dataset upload interface". The backend returns "recognitionTaskID".

[0033] 6. Automatic identification of multiple attributes of the pipeline network Model invocation: The FLASK backend forwards the dataset to the recognition model through the "Model Multi-Attribute Recognition Interface" (POST request, carrying referenceSize and photosURL parameters); Attribute Output: Based on the reference object size and photo features, the model calculates and returns four core data categories: Pipe width: accurate to 0.1cm (e.g., "30.5cm"); Pipe height above ground: accurate to 0.1cm (e.g., "120.3cm"); Pipe crossover status: Boolean value ("exists" or "does not exist"); Cross count: an integer (e.g., "2", returned only if the cross condition is "exist"); Results Display: After the FLASK backend verifies that there are no missing data, it synchronizes the data to the APP through the "Recognition Result Query Interface" (GET request, carrying token and recognitionTaskID). The APP displays the data in a list format for construction personnel to confirm.

[0034] 7. Building the archived dataset Data Integration: The app automatically collects the following information and integrates it with the recognition results: Location data: The longitude and latitude of the current collection point are obtained through a GPS module (accuracy ≤ 10 meters); Time data: Get the current system time (accurate to the second, such as "2025-10-20 15:30:45"); Personnel data: Information about the currently logged-in operator's account (e.g., "user_001"); Attachment data: URLs of 4 compliant photos (stored on the MinIO file server, valid indefinitely); Format Encapsulation: Data is encapsulated into JSON format according to the requirements of the archiving system, containing three sub-objects: basic data collection information, pipeline attribute information, and attachment information. For example: "Collection Basic Information":{"Collection ID":"C001","User ID":"user_001","Project ID":"P005","GPS Longitude":"116.397","GPS Latitude":"39.908","Collection Time":"2025-10-20 15:30:45"}, "Pipeline Attribute Information":{"Pipe Width":"30.5cm","Pipe Height from Ground":"120.3cm","Pipe Crossing Status":"Exists","Number of Crossings":"2"}, "Attachment Information":{"Panoramic Photo URL":"xxx","Side View 1 URL":"xxx","Side View 2 URL":"xxx","Close-up Photo URL":"xxx"}.

[0035] 8. Automatic transmission and archiving feedback Data submission: After the construction personnel confirm that the data is correct, they click "Submit Archive". The APP transmits the JSON data to the FLASK backend through the "Archive Submission Interface" (POST request, carrying token and collectionData parameters); File writing: The FLASK backend transmits data to the construction unit's file system through the "file data writing interface" (POST request, carrying collectionBase, pipeAttr, and photoUrls parameters); Results feedback: After receiving the data, the archive system automatically assigns an archive number (such as "SG-GW-2025-089"), writes it into the "Construction Pipeline Location Information Database", and returns "archiveResult: Success" and "archiveNo:SG-GW-2025-089". Process complete: The FLASK backend synchronizes the archiving results and file number to the APP. The APP notifies the construction personnel via voice ("Archiving successful, file number SG-GW-2025-089") and pop-up window, and the entire data collection and reporting process is completed.

[0036] Example 1: Taking the on-site data collection of a municipal water supply and drainage pipeline construction project (Project ID: P005) as an example, the specific implementation process is as follows: Preliminary preparations: The construction workers installed the "Construction Pipeline Information Collection APP" on their mobile phones (Android 12.0 system), deployed the FLASK backend and recognition model on the cloud server (CentOS 7.9, 4 cores 8G), prepared a basketball (diameter 24.6cm) as a reference object, and completed the interface adaptation between the FLASK backend and the project file system. On-site login: Construction personnel log in to the APP via WeChat authorization. The FLASK backend verifies that their project is P005 and grants them access to data collection. Placement of reference point: Place the basketball 1 meter to the right of the water supply and drainage pipe trench, ensuring no obstruction; Photo taking: Follow the APP voice prompts to take 4 photos (panoramic photo including construction site fence, side photo including basketball and pipe trench, close-up photo including pipe joints). Verification and Recognition: After the photo is uploaded, the recognition model passes the verification. The APP selects "basketball" as the reference object, and the model returns the pipe width as 32.2cm, the height from the ground as 118.5cm, the pipe intersection as "exists", and the number of intersections as 1. Archiving complete: The APP automatically obtains the GPS coordinates (longitude 116.402, latitude 39.915) and the collection time 2025-10-2016:15:30, constructs and submits the archived data, and the archive system returns the number "SG-GW-2025-092". The APP prompts that the archiving was successful.

[0037] This application employs a standardized data collection process involving voice-guided shooting, automatic verification, and direct connection to the archival system. This eliminates manual measurement, recording, and format conversion, reducing single-point collection time from the traditional 30 minutes to within 5 minutes, significantly improving efficiency. A multimodal, multi-attribute identification model for construction pipelines is used, based on known-size reference objects. Pipeline size identification error is ≤0.5cm, far lower than the 5-10cm error of manual estimation, improving data accuracy and avoiding the problem of missing key information in photos, facilitating later archival traceability. Multi-attribute data records are constructed with a unified format for easy interaction. A multi-level access control system with project binding and targeted uploads, along with encrypted data storage using user passwords and unique, tamper-proof photo URLs, prevents data leakage or misoperation, ensuring high security. It is highly scalable, supporting the addition of new reference object types and expanded identification attributes to adapt to the collection needs of different municipal pipeline networks.

[0038] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

Claims

1. A method for collecting and reporting multi-attribute information of municipal engineering pipeline networks based on a large model, characterized in that, Using the data acquisition app, which has a built-in multimodal driven construction pipeline multi-attribute recognition large model, the steps include: S1. Open and log in to the data collection app; S2. Reference object setting: Place the reference object according to the voice prompts and match the reference object size parameters; S3. Standardized photo shooting: Take photos in a fixed order according to voice prompts and upload them; S4. Multimodal dataset construction: Construct a multimodal dataset in the format of "reference object size + compliant photo URL + user ID + project ID", transmit it to the backend, and the backend returns "recognition task ID"; S5. Automatic identification of multiple attributes of pipeline network, multimodal driven construction pipeline multi-attribute identification large model is based on the size of reference object and photo features, calculates and returns pipeline network attribute information, with no missing information, and uploads it; S6. Constructing the archived dataset: The APP automatically collects location data, time data, personnel data, and attachment data. The data is encapsulated into JSON format according to the requirements of the archive system. The encapsulated data includes three sub-objects: basic collection information, pipeline attribute information, and attachment information.

2. The method for collecting and reporting multi-attribute information of municipal engineering pipeline networks based on a large model according to claim 1, characterized in that, It also includes the deployment and configuration of the data collection app, the deployment and interface configuration of the cloud recognition model, and the adaptation of the archive system interface.

3. The method for collecting and reporting multi-attribute information of municipal engineering pipeline networks based on a large model according to claim 2, characterized in that, The deployment and configuration of the data collection app includes: GPS permission management, intelligent photo processing, and targeted upload control.

4. The method for collecting and reporting multi-attribute information of municipal engineering pipeline networks based on a large model according to claim 3, characterized in that, GPS Permission Management: When the APP is launched for the first time, a pop-up window requests location permission to ensure that the latitude and longitude accuracy of the collection point is ≤10 meters, providing a basis for pipeline location tracing.

5. The method for collecting and reporting multi-attribute information of municipal engineering pipeline networks based on a large model according to claim 3, characterized in that, Intelligent photo processing: Automatically adapts to the phone's camera resolution, compresses photos to 1-2MB after shooting, and limits the format to JPG / PNG.

6. The method for collecting and reporting multi-attribute information of municipal engineering pipeline networks based on a large model according to claim 3, characterized in that, Targeted upload control: The app only transmits data to the pre-defined FLASK backend dedicated interface and prohibits sending data to third-party addresses.

7. The method for collecting and reporting multi-attribute information of municipal engineering pipeline networks based on a large model according to claim 1, characterized in that, Reference point location requirements: The reference point must be placed on the same horizontal side as the laid pipeline, at a distance of ≤1 meter from the pipeline, to ensure that the pipeline and the reference point can be captured simultaneously when taking the photo.

8. The method for collecting and reporting multi-attribute information of municipal engineering pipeline networks based on a large model according to claim 1, characterized in that, Standardized photographs include: panoramic photos, side views, and close-up photos.

9. The method for collecting and reporting multi-attribute information of municipal engineering pipeline networks based on a large model according to claim 1, characterized in that, The pipeline network attribute information includes: pipeline width, pipeline height above ground, pipeline intersections, and number of intersections.

10. The method for collecting and reporting multi-attribute information of municipal engineering pipeline networks based on a large model according to claim 1, characterized in that, The data collection app obtains the longitude and latitude of the current collection point through the GPS module to obtain location data; the data collection app obtains the current system time as time data; the data collection app obtains the currently logged-in operator's account information as personnel data; and the data collection app uses the URL of compliant photos as attachment data.

Images