Property project work order circulation control method and device, equipment and medium

Abstract

The invention relates to a property project work order circulation control method and device, equipment and medium in the technical field of property information, and the method comprises the steps: responding to a property project establishment event, and generating a project description text of a property project according to a result of recognizing a corresponding property current situation map, the information comprises an abnormal problem existing in a property site in the graph and description of processing indication of the abnormal problem; after a project work order of the property project is created based on the project description text, the work order is sent to an adaptive employee; in response to a project feedback event of the property project reported by the proper employee, a progress description text of the property project is generated according to a result of identifying the corresponding property operation diagram, and the progress description text comprises the project operation information of the property site in the diagram and description of the project operation progress; and generating a project progress report based on the project work order and the progress description text thereof, and pushing the project progress report to a preset interface. Property project work order circulation can be automatically controlled, and it is ensured that the execution process is accurate, reliable and efficient.

Description

Technical Field

[0001] This application relates to the field of property information technology, and in particular, to a method for controlling the flow of work orders for property projects, as well as corresponding devices, computer equipment, and computer-readable storage media. Background Art

[0002] In the field of property management, the daily work cannot do without the recording of pictures of the work scene. However, in the field of property management, the application of image recognition technology is still in its infancy. The existing property management systems are weak in the automated processing of picture-recorded information, unable to give full play to the advantages of image recognition technology, restricting the intelligent development of property management, and resulting in the inability to serve the intelligent advancement of the property management process.

[0003] In view of the deficiencies of traditional technologies, the applicant has made corresponding explorations. Summary of the Invention

[0004] The primary objective of this application is to solve at least one of the above problems and provide a method for controlling the flow of work orders for property projects, as well as corresponding devices, computer equipment, and computer-readable storage media.

[0005] To meet the various objectives of this application, the following technical solutions are adopted:

[0006] A method for controlling the flow of work orders for property projects provided to meet one of the objectives of this application includes the following steps:

[0007] In response to a property project establishment event, call an image processing algorithm to identify the image information of the property current situation map corresponding to this event, and generate a project description text for the property project according to the recognition result. The project description text includes a description of the abnormal problems existing in the property site in this map and their handling instructions;

[0008] Adopt a language inference model to create a project work order for the property project based on the project description text, and dispatch the project work order to a suitable and competent employee;

[0009] In response to the suitable and competent employee reporting a project feedback event for the property project, call an image processing algorithm to identify the image information of the property operation map corresponding to this event, and generate a progress description text for the property project according to the recognition result. The progress description text includes a description of the project operation information and the project operation progress at the property site in this map;

[0010] Adopt a language inference model to generate a project progress report based on the project work order and its progress description text, and push it to a preset interface.

[0011] On the other hand, a property project work order transfer control device provided to meet one of the purposes of this application includes a first event response module, a project work order dispatching module, a second event response module, and a progress report pushing module. Among them, the first event response module is used to respond to the property project establishment event, call an image processing algorithm to identify the image information of the property current situation map corresponding to this event, and generate a project description text of the property project according to the recognition result. The project description text includes the description of the abnormal problems existing in the property site in this map and their handling instructions; the project work order dispatching module is used to create the project work order of the property project based on the project description text by using a language inference model, and dispatch the project work order to a suitable and competent employee; the second event response module is used to respond to the project feedback event of the property project reported by the suitable and competent employee, call an image processing algorithm to identify the image information of the property operation map corresponding to this event, and generate a progress description text of the property project according to the recognition result. The progress description text includes the description of the project operation information and the project operation progress at the property site in this map; the progress report pushing module is used to generate a project progress report based on the project work order and its progress description text by using a language inference model, and push it to a preset interface.

[0012] On the other hand, a computer device provided to meet one of the purposes of this application includes a central processing unit and a memory. The central processing unit is used to call and run a computer program stored in the memory to execute the steps of the property project work order transfer control method described in this application.

[0013] On the other hand, a computer-readable storage medium provided to meet another purpose of this application stores a computer program implemented according to the property project work order transfer control method in the form of computer-readable instructions. When the computer program is called and run by a computer, it executes the steps included in this method.

[0014] The technical solution of this application has many advantages, including but not limited to the following aspects:

[0015] First of all, in this application, when a property project is initiated, an image processing algorithm is called to identify the image information of the current property map, accurately extract the actual situation of the property site in the map, including existing abnormal problems and their handling instructions. Then, based on these key information, a project work order is generated and assigned to a suitable employee for execution. Furthermore, when the project feedback is received, an image processing algorithm is called to identify the image information of the current property map, accurately extract the project operation information and progress in the map. Then, based on these key information, a project report is generated and pushed to a preset interface for subsequent processing. It can be seen that it is possible to achieve full-process automation and intelligent control from the initiation of a property project to the generation of work orders, dispatching, progress feedback, and progress report pushing, greatly improving the processing ability of picture record information in property management work, ensuring the accuracy, reliability, and efficiency of the entire work order circulation control process, being able to promptly discover and handle abnormal problems occurring at the property site, facilitating the supervision of the specific project progress, and moreover, without the need for users or employees to manually edit work orders and progress reports from scratch, reducing omissions, lowering the threshold and time of manual editing, improving the document output efficiency, and significantly enhancing the user experience. Brief Description of the Drawings

[0016] The above and / or additional aspects and advantages of this application will become apparent and be easily understood from the following description of the embodiments in conjunction with the drawings, where:

[0017] Figure 1 is a flowchart of a typical embodiment of the property project work order circulation control method of this application;

[0018] Figure 2 is a principle block diagram of the property project work order circulation control device of this application;

[0019] Figure 3 is a structural diagram of a computer device adopted by this application. Detailed Embodiments

[0020] The embodiments of this application are described in detail below. The examples of the embodiments are shown in the drawings, where the same or similar reference numerals represent the same or similar elements or elements with the same or similar functions throughout. The embodiments described below with reference to the drawings are exemplary and are only used to explain this application and should not be construed as a limitation of this application.

[0021] Those skilled in the art can understand that, unless specifically stated otherwise, the singular forms "a", "an", "the" and "said" used herein may also include the plural forms. It should be further understood that the term "comprising" used in the specification of the present application means the presence of the stated features, integers, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or their groups. It should be understood that when we say that an element is "connected" or "coupled" to another element, it can be directly connected or coupled to other elements, or there may also be intermediate elements. In addition, the "connection" or "coupling" used herein may include wireless connection or wireless coupling. The phrase "and / or" used herein includes all or any unit and all combinations of one or more of the associated listed items.

[0022] Those skilled in the art can understand that, unless otherwise defined, all terms used herein (including technical terms and scientific terms) have the same meaning as the general understanding of those of ordinary skill in the art to which the present application belongs. It should also be understood that terms such as those defined in a general dictionary should be understood as having a meaning consistent with the meaning in the context of the prior art, and will not be interpreted in an idealized or overly formal sense unless specifically defined as here.

[0023] Those skilled in the art can understand that the "client", "terminal", and "terminal device" used herein include both devices with wireless signal receivers that only have the ability to receive and no ability to transmit, and devices with receiving and transmitting hardware that have the receiving and transmitting hardware capable of two-way communication on a two-way communication link. Such devices can include: cellular or other communication devices such as personal computers, tablet computers, etc., which have a single-line display or a multi-line display or a cellular or other communication device without a multi-line display; PCS (Personal Communications Service), which can combine voice, data processing, fax, and / or data communication capabilities; PDA (Personal Digital Assistant), which can include a radio frequency receiver, pager, Internet / intranet access, web browser, notepad, calendar, and / or GPS (Global Positioning System) receiver; conventional laptop and / or palm-top computers or other devices, which are conventional laptop and / or palm-top computers or other devices with and / or including a radio frequency receiver. The "client", "terminal", and "terminal device" used herein can be portable, transportable, installed in a vehicle (air, sea, and / or land), or suitable for and / or configured to run locally, and / or run in a distributed form at any other location on the earth and / or in space. The "client", "terminal", and "terminal device" used herein can also be a communication terminal, an Internet access terminal, a music / video playback terminal, for example, it can be a PDA, MID (Mobile Internet Device), and / or a mobile phone with music / video playback function, or it can also be a smart TV, a set-top box, etc.

[0024] The hardware referred to by names such as "server", "client", and "service node" in this application is essentially an electronic device with the equivalent capabilities of a personal computer, and is a hardware device with the necessary components disclosed by the von Neumann principle, including a central processing unit (including an arithmetic unit and a controller), a memory, an input device, and an output device. The computer program is stored in its memory, and the central processing unit loads the program stored in the external memory into the memory for execution, executes the instructions in the program, and interacts with the input / output devices to complete specific functions.

[0025] It should be noted that the concept of "server" in this application can similarly be extended to the case of server clusters. According to the network deployment principles understood by those skilled in the art, the various servers should be logically divided. Physically, these servers can either be independent of each other but can be invoked through interfaces, or be integrated into a physical computer or a set of computer clusters. Those skilled in the art should understand this flexibility and should not be restricted by this when implementing the network deployment method of this application.

[0026] One or several technical features of this application, unless explicitly specified, can either be deployed on the server and accessed by the client remotely invoking the online service interface provided by the server, or be directly deployed and run on the client for access.

[0027] The neural network models cited or possibly cited in this application, unless explicitly specified, can either be deployed on a remote server and remotely invoked on the client, or be deployed on a client with sufficient device capabilities for direct invocation. In some embodiments, when it runs on the client, its corresponding intelligence can be obtained through transfer learning to reduce the requirements for the client's hardware operation resources and avoid excessive consumption of the client's hardware operation resources.

[0028] All kinds of data involved in this application, unless explicitly specified, can either be remotely stored on the server or stored on the local terminal device, as long as it is suitable for being invoked by the technical solution of this application.

[0029] Those skilled in the art should be aware that although the various methods of this application are described based on the same concept and thus show commonality with each other, unless otherwise specified, these methods can all be executed independently. Similarly, for the various embodiments disclosed in this application, they are all proposed based on the same inventive concept. Therefore, for concepts with the same expression, as well as concepts that are only appropriately transformed for convenience although the concept expressions are different, they should be equivalently understood.

[0030] For the various embodiments to be disclosed in this application, unless explicitly pointed out that there is a mutually exclusive relationship between them, the relevant technical features involved in each embodiment can be cross-combined to flexibly construct new embodiments, as long as this combination does not deviate from the creative spirit of this application and can meet the requirements in the prior art or solve certain deficiencies in the prior art. Those skilled in the art should be aware of this flexibility.

[0031] A method for controlling the workflow of work orders for property projects in this application can be programmed as a computer program product and implemented by running on a client or a server. For example, in an exemplary application scenario of this application, it can be implemented by deploying it on the server of an e-commerce platform. Thus, by accessing the interface opened after the computer program product runs, human-computer interaction can be performed with the process of the computer program product through a graphical user interface to execute this method.

[0032] Please refer to Figure 1 , in a typical embodiment of the method for controlling the workflow of work orders for property projects in this application, it includes the following steps:

[0033] Step S1100: Respond to the property project establishment event, call an image processing algorithm to identify the image information of the property current situation map corresponding to this event, and generate a project description text for the property project according to the recognition result. The project description text includes descriptions of abnormal problems existing in the property site in this map and their handling instructions;

[0034] This application builds and maintains a property system, and both property owners and employees of various properties can log in to use the corresponding services provided by this system. The employees of various properties include maintenance employees, cleaning employees, customer service personnel, greening maintenance employees, administrative employees, customer service employees, administrators, inspection employees, etc.

[0035] In the property management system, work orders are usually initiated by property owners or inspection personnel. When these users discover abnormal problems at the property site, they can log in to the system, enter the graphical user interface, and click the "Create Work Order" function control to trigger the property project establishment event. After the system responds to this event, it will start the process of creating a work order for the user. During this process, the user can click the "Take Photo" function control on this interface to enable the camera unit in their terminal device to take an on-site photo of the actual situation at the property site, thus obtaining a property current situation map to be watermarked. Subsequently, the user uploads this map to the system. The system obtains the timestamp corresponding to the execution of the shooting action by the terminal device, translated into the shooting time, with the user's permission, and obtains the longitude and latitude corresponding to the execution of the shooting action by the terminal device to locate the administrative division address on the map, and sends the shooting time and the administrative division address to the user for confirmation. The user needs to confirm the shooting time and append the property division address of the property site to the administrative division address to obtain the property site location information, and then submit the confirmed shooting time and the property site location information to the system. Further, the system generates a watermark image containing the shooting time and the property site location information, and superimposes the watermark image on the layer of the property current situation map to be watermarked according to the preset position in the property current situation map, completing the watermark addition operation to obtain the property current situation map. The superimposed position of the preset watermark pattern can be set as needed by those skilled in the art. For ease of understanding, a demonstrative example is: "Shooting time: Friday, January 12, 2024, 11:08, Property site location information: Corridor on the Xth floor of Building X in XXX Community, Qiwen Road, Haishu District, Ningbo City, Zhejiang Province."

[0036] In one embodiment, the following steps are included: Step S1110, perform object detection on the property current situation map corresponding to the property project establishment event using a preset object detection model, determine the entity status labels corresponding to each object detection target in the map, and determine the abnormal problems corresponding to the entity status labels;

[0037] The object detection model is pre-called and fine-tuned using a first training set until it converges, acquiring the ability to determine each object detection target and its entity status label in the input image. Those skilled in the art can flexibly implement the model training here. Thus, taking the property current situation map as the input, the object detection model determines each object detection target in the map and obtains the entity status label corresponding to each object detection target. Then, it traverses each entity status label to determine whether there is a normal or abnormal string in the entity status label, and uses a delimiter to splice each entity status label with an abnormal string into an abnormal problem. The delimiter can be a space, a comma, a semicolon, etc.

[0038] The target detection model can be any pre-trained model suitable for target detection tasks in the field of image processing, such as YOLO, U-Net, DeepLab series, BiSeNet, etc. The model has learned in advance the ability to recognize various real objects in the input image.

[0039] Prepare the first training set. Specifically, take photos of multiple different property sites in advance, obtain the corresponding unwatermarked property status pictures as the first training samples, and then regard all actual objects related to property work in any property status picture as target detection objects. Property work includes inspection, cleaning, maintenance, greening maintenance, etc. For each property status picture, use labelme and labelimg marking tools to frame all target detection objects in the picture. For each target detection object, first determine what the target detection object is, the status of the target detection object, and judge whether the status is normal. Then, edit the text describing the corresponding actual object, the status of the actual object and the judgment result of the status as the entity status label. It can be understood that it is the first supervision label corresponding to the property status picture as the first training sample. For ease of understanding, an exemplary example is given: "abnormal cracked wall", "normal wall", "abnormal garbage ground", "normal ground", "abnormal disorderly placed fire equipment". The first training set is composed of the first training samples and their first supervision labels.

[0040] Step S1120: classify the property status map using a preset project classification model to obtain a project type text;

[0041] The project classification model is pre-trained with the second training set until convergence, and acquires the ability to determine the project type to which the input image belongs. Those skilled in the art can flexibly implement the model training here. Thus, the project classification model takes the property status map as input, determines the project type to which the map belongs, and outputs the project type text in text form.

[0042] The project classification model can be any model suitable for image classification tasks in the field of image processing, such as ResNet, VGG, Inception series, MobileNet, etc.

[0043] Prepare the second training set. Specifically, collect multiple property status pictures of different project types in advance as second training samples. Then, for each property status picture, determine the project type to which the picture belongs according to the actual scene and property work content or object in the picture, such as "inspection project", "cleaning project", "maintenance project", "greening maintenance project", etc., and use the project type text as the second supervisory label corresponding to the property status picture as the second training sample. Aggregate each of the second training samples and their second supervisory labels to form the second training set.

[0044] Step S1130: using an optical character recognition algorithm to perform character recognition on the watermark image in the property status map to obtain watermark text information, wherein the watermark information includes shooting time and property site location information;

[0045] Optical Character Recognition (OCR) is a technology that can extract text information from images. In the property status map, the watermark image contains the shooting time and property location information. The OCR algorithm can realize automatic recognition and extraction of this information.

[0046] First, the OCR algorithm will perform image preprocessing on the property status map, including grayscale, binarization, noise removal and other operations to improve the contrast and clarity of the image and enhance the recognizability of the text part. Next, the algorithm will perform text detection and determine the area in the image that may contain text by analyzing features such as edges, contours and shapes in the image. This step is usually implemented using a sliding window-based method or a deep learning-based text detection model.

[0047] After text detection is completed, the OCR algorithm enters the character recognition stage. This stage segments the detected text area into individual characters or character sequences, and then converts the characters in the image into corresponding text information by matching them with a predefined character set. Deep learning models such as convolutional neural networks (CNNs) and recurrent neural networks (RNNs) are usually used to improve the accuracy of character recognition.

[0048] Finally, the OCR algorithm will perform post-processing of the results, including correction and verification of the recognized text to ensure the accuracy and completeness of the extracted watermark text information. Through the above steps, the optical character recognition algorithm can effectively extract the shooting time and property location information from the watermark image in the property status map, providing accurate data support for property management.

[0049] Step S1140: After guiding the language reasoning model to infer the processing instruction text of the abnormal problem according to the abnormal problem text, the project type text, and the watermark text information, a corresponding project description text is generated.

[0050] Language reasoning models can be selected from natural language processing models such as GPT4 and Deepseek, which are pre-trained to a convergent state and have strong understanding and reasoning capabilities. For example, the Deepseek model also has strong capabilities in understanding, reasoning and expression, and can effectively integrate and analyze complex information.

[0051] The abnormal problem text, project type text, and watermark text information are embedded in the preset first prompt template to obtain the first prompt text. The first prompt text is input into the language reasoning model to obtain the project description text output by the model. In one embodiment, the first prompt template is: "Please infer the processing instruction text of the abnormal problem from the perspective of property management based on the abnormal problem, project type text, and watermark text information provided below. It should be clear and specific, indicating the actual processing work required. Finally, after obtaining the processing instruction text, the corresponding project description text is generated by combining the abnormal problem, project type text, watermark text information, and processing instruction text. The project description text should clearly and completely describe the overall situation of the property project. Abnormal problem: {abnormal problem to be embedded}; project type text: {project type text to be embedded}, watermark text information: {watermark text information to be embedded}.".

[0052] Step S1200: using a language reasoning model to create a project work order for the property project based on the project description text, and assigning the project work order to a suitable employee;

[0053] Embed the project description text into the preset second prompt template to obtain the second prompt text. Input the second prompt text into the language reasoning model to obtain the project work order output by the model. At this point, the work order creation process is completed and ends. In one embodiment, the second prompt template is: "You are a work order intelligent assistant and can generate a structured and complete project work order based on the following project description content. The project work order should include project type, project location, task description, and project establishment time. Project description content: {project description text to be embedded}.".

[0054] In one embodiment, a preset dual-tower model is used, and the task description text in the project work order and the function description text of each employee in the property employee database are respectively formed into a single text pair, which is input into the dual-tower model. For each text pair, the two text feature towers in the model respectively extract the single text in the text pair, and respectively extract the deep semantic information of each text, and map the deep semantic information to the same semantic space to obtain a vectorized representation of the text feature vector corresponding to each deep semantic information. Then, the matching output layer in the model calculates the vector similarity between the two text feature vectors as the matching degree and outputs it.

[0055] Obtain the matching degree between the task description text and the function description text output by the model. Furthermore, employees with a relatively high matching degree can be screened out as candidate suitable employees. Associate the project work orders with these employees and their function description texts and send them to the administrator through the property system. The administrator selects the suitable employees. Thus, the project work orders are bound to the suitable employees. A work order interface can be created and displayed in the system, and all project work orders are displayed in this interface so that suitable employees can select the project work orders suitable for themselves to handle in this interface and take over the handling of the work orders. In a flexible business scenario, a suitable employee can refuse to accept a project work order suitable for themselves. Thus, the rejected project work order can be taken over by other employees. After taking over, the employee who takes over and their function description text are sent to the administrator. After the administrator's review and approval, the employee is confirmed as the suitable employee for the project work order and is responsible for handling the project work order.

[0056] The dual tower model has been pre-trained to a convergent state and has acquired the ability to determine the matching degree between the input function description text and the task description text. Those skilled in the art can flexibly implement the model training here. The model structure of the dual tower model is two identical text feature representation towers with shared parameters, followed by a matching degree output layer after the two text feature representation towers. The text feature representation tower is suitable for extracting the semantics of the input text for vector representation, and can be selected from a variety of known models, including but not limited to any one of Bert, RNN, BiLSTM, BiGRU, RoBERTa, ALBert, ERNIE, BERT-WWM, etc. The matching degree output layer can be implemented by any one of the cosine similarity algorithm, vector dot product algorithm, Manhattan distance, Euclidean distance algorithm, Pearson correlation coefficient, etc.

[0057] Step S1300: In response to the suitable employee reporting the project feedback event of the property project, call an image processing algorithm to identify the image information of the property operation diagram corresponding to the event, and generate a progress description text of the property project according to the recognition result. The progress description text includes the project operation information and the description of the project operation progress at the property site in the diagram;

[0058] When a qualified employee takes over the processing of project work orders, they need to be responsible for the property project and report the corresponding project operation progress. In this regard, the qualified employee can specify a project work order on the system and then perform a reporting operation on it. For example, a "Report" function control can be provided on the details interface of the project work order for the user to click. As a result, the system responds to the project feedback event and starts the process of creating a project progress report for the qualified employee. During this process, the project progress report interface is displayed to the qualified employee. The qualified employee enables the camera unit in their terminal device by clicking the "Take Photo" function control on this interface, and takes an on-site photo of the property operation situation, thus obtaining a property operation map to be watermarked. Subsequently, the qualified employee uploads this map to the system. The system obtains the time stamp corresponding to the execution of the shooting behavior by the terminal device and translates it into the shooting time with the permission of the qualified employee, and locates the administrative division address on the map based on the longitude and latitude corresponding to the execution of the shooting behavior by the terminal device, and sends the shooting time and the administrative division address to the qualified employee for confirmation. The qualified employee needs to confirm the shooting time and append the property division address at the property site to the administrative division address to obtain the property site location information, and then submit the confirmed shooting time and the property site location information to the system. Further, the system generates a watermark image containing the shooting time and the property site location information, and superimposes the watermark image on the layer of the property operation map to be watermarked according to the preset position in the property operation map to be watermarked, and completes the watermarking operation to obtain the property operation map. The superimposed position of the preset watermark pattern can be set as needed by those skilled in the art.

[0059] In one embodiment, the following steps are included: Step S1310, perform object detection on the property operation map corresponding to the project feedback event using a preset target detection model, determine the entity status labels corresponding to all target objects (target detection objects) in the map, and summarize all entity status labels to obtain a current situation detail text;

[0060] The target detection model can take the property operation map as input, determine each target detection object in the map, obtain the entity status label corresponding to each target detection object, and splice each entity status label into a current situation detail text using a delimiter.

[0061] Step S1320, classify the property operation map using a preset progress classification model to determine the corresponding progress type text;

[0062] The progress classification model is pre-called and trained on a third training set until it converges, and acquires the ability to determine the progress type to which the input picture belongs. Those skilled in the art can flexibly implement the model training here. Thus, the progress classification model takes the property operation map as input, determines the progress type to which the map belongs, and outputs the progress type text in text form.

[0063] The progress classification model can be any model suitable for image classification tasks in the field of image processing, such as ResNet, VGG, Inception series, MobileNet, etc.

[0064] Prepare the third training set. Specifically, collect multiple property operation drawings of different progress types in advance as third training samples. Then, for each property operation drawing, determine whether the progress type of the drawing is "project completed" or "project not completed" according to the operation situation in the drawing, and use the progress type text as the third supervisory label corresponding to the property operation drawing as the third training sample. Aggregate the third training samples and their third supervisory labels to form the third training set.

[0065] Step S1330: Use an optical character recognition algorithm to perform character recognition on the watermark image in the property operation map to obtain watermark text information, where the watermark information includes shooting time and property site location information;

[0066] Optical Character Recognition (OCR) is a technology that can extract text information from images. In the property operation map, the watermark image contains the shooting time and property site location information. The OCR algorithm can realize automatic recognition and extraction of this information.

[0067] First, the OCR algorithm will perform image preprocessing on the property operation map, including grayscale, binarization, noise removal and other operations to improve the contrast and clarity of the image and enhance the recognizability of the text part. Next, the algorithm will perform text detection and determine the area in the image that may contain text by analyzing features such as edges, contours and shapes in the image. This step is usually implemented using a sliding window-based method or a deep learning-based text detection model.

[0068] After text detection is completed, the OCR algorithm enters the character recognition stage. This stage segments the detected text area into individual characters or character sequences, and then converts the characters in the image into corresponding text information by matching them with a predefined character set. Deep learning models such as convolutional neural networks (CNNs) and recurrent neural networks (RNNs) are usually used to improve the accuracy of character recognition.

[0069] Finally, the OCR algorithm will perform post-processing of the results, including correction and verification of the recognized text to ensure the accuracy and completeness of the extracted watermark text information. Through the above steps, the optical character recognition algorithm can effectively extract the shooting time and property site location information from the watermark image in the property operation map, providing accurate data support for property management.

[0070] Step S1340: Use a language inference model to generate a progress description text based on the current situation details text, progress type text, and watermark text information.

[0071] Embed the abnormal problem text, project type text, and watermark text information into a preset third prompt template to obtain a third prompt text. Input the third prompt text into the language inference model to obtain the project description text output by the model. In one embodiment, the third prompt template is: "Please generate a progress description text from the perspective of property management based on the following provided current situation details text, progress type text, and watermark text information. The progress description text should clearly and completely describe the operation situation of the property project and the status of the project operation progress. It should include the description of project operation information and project operation progress. The provided current situation details text: {to-be-embedded current situation details text}, progress type text: {to-be-embedded progress type text}, watermark text information: {to-be-embedded watermark text information}."

[0072] To make it clearer for you about the above generation requirements and the generation effects to be achieved, here is an example of a generated progress description text for demonstration:

[0073] Demonstration example {Current situation details text in the example: Normal walls, abnormal ground with garbage;

[0074] Progress type text in the example: Project not completed;

[0075] Watermark text information in the example: Shooting time: Monday, January 15, 2024, 14:30, Property site location information: Corridor on the Xth floor of Building X in XXX Community, Qiwen Road, Haishu District, Ningbo City, Zhejiang Province;

[0076] Progress description text in the example: As of January 15, 2024, the property operation progress of the corridor on the Xth floor of Building X in XXX Community is as follows: The wall repair work has been completed, and the ground cleaning work generated by the repair is in progress. Currently, the overall project progress has not been completed.}

[0077] Step S1400: Use a language inference model to generate a project progress report based on the project work order and its progress description text, and push it to a preset interface.

[0078] The project work order and its progress description text are embedded in the preset fourth prompt template to obtain the fourth prompt text. The fourth prompt text is input into the language reasoning model to obtain the project progress report output by the model. At this point, the process of creating the project progress report is completed and ended. In one embodiment, the fourth prompt template is: "Please analyze the task description text in the project work order based on the project work order and its progress description text provided below, combine the progress description text to clarify the detailed execution status of all work content, and finally generate a formal project progress report. The project progress report should clearly and completely describe the overall progress of the property project."

[0079] The project progress report is pushed to a preset interface, and the preset interface further processes the project progress report according to pre-specified business requirements. Those skilled in the art can implement the preset interface in advance according to business requirements. For example, the business requirement may be to package the project progress report into a notification message, and then send the notification message to the administrator to inform the administrator so that the administrator can learn about the specific project progress; or the project progress report may be associated with a project work order for storage, and the system automatically updates the progress of the project work order.

[0080] It can be known from the typical embodiments of the present application that the technical solution of the present application has many advantages, including but not limited to the following aspects:

[0081] First, in this application, when the property project is established, the image processing algorithm is called to identify the image information of the property status map, accurately extract the actual situation of the property site in the map, including the existing abnormal problems and their processing instructions, and then generate a project work order based on these key information, and then assign the project work order to the appropriate employee for execution. Moreover, when the project is fed back, the image processing algorithm is called to identify the image information of the property status map, accurately extract the project operation information and progress in the map, and then generate a project report based on these key information, and push it to the preset interface for subsequent processing. It can be seen that the whole process from the establishment of the property project to the generation of work orders, dispatching, progress feedback and progress report push can be automated and intelligently controlled, which greatly improves the processing ability of image record information in property management work, ensures that the entire work order flow control process is accurate, reliable and efficient, can timely discover and handle abnormal problems on the property site, and facilitate the supervision of specific project progress, and does not require users or employees to personally edit work orders and progress reports from scratch, reducing omissions, and also reducing the threshold and time of manual editing, improving document output efficiency, and greatly improving user experience.

[0082] In a further embodiment, before step S1200, assigning the project work order to a suitable qualified employee, the following steps are included:

[0083] Step S2200: Match the task description text in the project work order with the function description texts of each employee in the property employee database to determine each candidate suitable employee;

[0084] In one embodiment, it is implemented by means of the RAG (Retrieval-Augmented Generation) technology combined with a language inference model. Specifically, first, a vector database of the property employee database is constructed. Index the function description texts of each employee in the property employee database, including cleaning and extracting the original data, and converting files in different formats into plain text. Then, use a text segmentation tool to segment these function description texts into small and manageable segments, that is, text blocks, for the language model to process. Next, apply an embedding model to encode each block of text and convert it into an abstract numerical representation in a high-dimensional vector space, thus abstracting the semantic content of the text and enabling it to be expressed and compared mathematically. All the generated text block vectors and their associated original text information are integrated into a vector database, which serves as a huge knowledge warehouse and can efficiently store and retrieve information.

[0085] When processing the project work order, take the task description text in the project work order as the input, and first convert it into a query vector through the same embedding model. In this way, the task description text is converted into a form that can be compared with the text block vectors stored in the vector database. Then, perform a similarity match between the query vector and all the stored vectors in the vector database. Through the vector similarity algorithm, find the top K text blocks that are closest to the query vector, that is, the fragments of the employee function description text that are most semantically relevant to the task description text. Then, take these relevant fragments of the employee function description text as context information and input them into the language inference model together with the task description text. The language inference model can combine this context information and the task description text to generate a comprehensive evaluation and analysis of the matching degree between the task and the employee functions, so as to determine each candidate suitable employee. It can be seen that not only makes full use of the powerful language understanding and generation capabilities of the language inference model, but also enhances the accuracy and reliability of the generation results through the retrieved external knowledge (employee function description text). Step S2210: Push the project work order associated with each candidate suitable employee and their function description texts to the property administrator;

[0086] The system generates a push notification or report that includes the detailed information of the project work order, the list of candidate suitable employees, and the functional description text of each candidate employee. This push notification or report can be displayed to the property manager through the interface of the property management system or sent to the property manager via email, instant message, etc. The pushed information can clearly show the requirements of the project work order and the functional matching of the candidate suitable employees, so that the property manager can quickly understand the suitability of each candidate employee. Step S2220: Obtain the suitable employees determined by the property manager and bind the suitable employees to the work order.

[0087] After receiving the push notification or report, the property manager can select one or more of the most suitable candidate suitable employees as the suitable employees according to the requirements of the project work order and the functional description text of the candidate suitable employees. The property manager can perform the selection operation through the interface of the property management system, such as checking the suitable employees in the list of candidate employees or selecting the suitable employees through a drop-down menu. Thus, the system can obtain the selection result of the property manager in real time and bind the selected suitable employees to the project work order. The binding operation can include adding the information of the suitable employees to the record of the project work order or establishing an association relationship between the suitable employees and the project work order in the database of the property management system. After the binding is completed, the system updates the status of the project work order, shows that the suitable employees have been determined, and notifies the relevant information to the suitable employees and the users in the relevant systems for subsequent work arrangements and executions.

[0088] In this embodiment, first, by using the RAG technology combined with the language inference model to match the task description text in the project work order with the functional description text of each employee in the property employee database, it is possible to accurately determine each candidate suitable employee. Compared with the traditional manual screening or simple keyword matching method, the accuracy and efficiency of the matching are greatly improved, and the possibility of manual intervention and errors is reduced. Secondly, pushing the project work order associated with each candidate suitable employee and their functional description text to the property manager provides comprehensive and intuitive information support for the property manager, enabling them to quickly understand the matching situation between the project work order and the employee functions, and thus make a correct decision quickly, improving the convenience and scientific nature of management. Finally, obtaining the suitable employees determined by the property manager and binding the suitable employees to the work order ensures that the project work order can be accurately assigned to the appropriate employees for processing, further improving the work efficiency and quality, and avoiding work delays caused by improper personnel arrangements.

[0089] In a further embodiment, after step S1200: dispatching the project work order to the suitable employees that match, the following steps are included:

[0090] Step S12001, responding to the event of a qualified employee accepting an order, obtaining the current face image and the project work order corresponding to the event, and determining the similarity between the current face image and the stored face image of the qualified employee in the property employee database;

[0091] The qualified employee can check and select the project work order suitable for him / her in the work order interface and take it over for processing. When the qualified employee confirms the project work order to be taken over, the system will be triggered to respond to the qualified employee's order event, and then the qualified employee needs to take a real-time photo of his / her face and upload the obtained real-time face image to the system. In this way, the system obtains the project work order and the real-time face image, and then pre-processes the real-time face image and the qualified employee's stored face image, including operations such as image grayscale and normalization, to reduce the noise and light effects of the image. Then, the open source pre-trained deep learning model for face recognition tasks in the field of image processing is used to extract the feature vector of the face image, and convert the face image into an abstract numerical representation in a high-dimensional feature space. Finally, the similarity between the two image feature vectors corresponding to the two images is calculated. The similarity calculation method can be any one of cosine similarity, Euclidean distance, etc. After pre-training to a convergence state, the pre-trained deep learning model learns the ability to extract and vectorize the face features in the input image.

[0092] Step S12002: When the similarity exceeds a preset threshold, the project work order is added to the work order list of the qualified employee.

[0093] A preset threshold is set in advance. When the similarity between the current face image and the stored face image exceeds the threshold, it means that the faces in the two images point to the same person, that is, the qualified employee. In this way, the identity verification of the qualified employee is passed. At this time, the system adds the project work order to the work order list of the qualified employee so that the qualified employee can view and process the project work order.

[0094] In this embodiment, first, identity verification is performed through face recognition technology. This whole process forms a reliable chain of evidence, which can prove that the designated qualified employee personally takes the order, effectively preventing unqualified employees from taking the order and qualified employees from not recognizing the order after taking the order, improving the security of project work order processing, and avoiding potential risks caused by identity fraud. Secondly, the automated face recognition and similarity comparison process significantly improves work efficiency, reduces the time and workload of identity verification, and enables project work orders to be distributed to qualified employees more quickly, thereby speeding up the operation of the entire workflow.

[0095] In a further embodiment, step S1300, calling an image processing algorithm to perform image information recognition on a property operation diagram corresponding to a project feedback event, and generating a progress description text of the property project according to the recognition result, comprises the following steps:

[0096] Step S1310: Retrieve the project detailed types in the quality inspection result library that match the processing instructions described in the project work order, and obtain each candidate inspection result text associated with the project detailed type;

[0097] The quality inspection result library contains multiple project detailed types and their associated inspection result texts. For each project detailed type, the associated inspection result texts are respectively used to describe the results of inspecting the associated project progress of the corresponding property site and its detailed operation conditions, including at least one detailed operation condition of the property site when the project is completed, and description texts corresponding to multiple detailed operation conditions of the property site when the project is not completed. The project detailed type is the behavioral description of the work done at the property site and the corresponding object description. Those skilled in the art can flexibly build the quality inspection result library according to the disclosure here.

[0098] For easy understanding, by way of example, the quality inspection result library contains a project with a project detailed type of "external wall cleaning". For this project detailed type, the associated inspection result texts include:

[0099] Inspection result text when the project is completed: "The project is completed. There are no obvious stains on the external wall surface. The boundary of the cleaning area is clear. There are no residual water stains or cleaner traces."; "The project is completed. The external wall cleaning work has been completed. The surface is smooth and there is no stain residue, meeting the cleaning standard."

[0100] Inspection result text when the project is not completed: "The project work is not completed. There are still obvious stains in some areas of the external wall. The cleaning work has not covered all areas."; "The project work is not completed. The external wall cleaning work is not completed. There are residual water stains in some areas and further cleaning is required."; "The project is not completed. The external wall cleaning work is not completed. The cleaner has not been rinsed clean in some areas and needs to be reprocessed.

[0101] In one embodiment, a dual - tower language model can be adopted to achieve an efficient and accurate matching between the processing instructions described in the project work order and the detailed project types in the quality inspection result library. Specifically: The dual - tower language model is a deep - learning model structure. Its core idea is to construct independent feature representation networks (two towers) for two entities (such as queries and documents) respectively, and match them through similarity calculation in a shared semantic space. Take the processing instructions described in the project work order as the query input, and the detailed project types in the quality inspection result library as the document input, and perform feature extraction through two independent neural networks (i.e., the two towers) respectively. Each tower encodes its input features separately to generate low - dimensional dense vectors (Embeddings), thus mapping the processing instructions and the detailed project types into the same semantic space. Then, through similarity calculation (such as cosine similarity or dot product), the matching score between the processing instructions and the detailed project types is obtained. Specifically, the calculation formula is:

[0102]

[0103] Among them, \(h_q\) represents the feature vector of the processing instructions, \(h_d\) represents the feature vector of the detailed project type, \(\|h_q\|\) represents the norm of \(h_q\), \(\|h_d\|\) represents the norm of \(h_d\), and \(s(q, d)\) represents the matching degree between the processing instructions and the detailed project type. Thus, the detailed project type with the highest matching degree and exceeding the preset threshold can be selected, and then each candidate detection text associated with this detailed project type in the quality inspection result library can be obtained. The preset threshold can be set by those skilled in the art as needed.

[0104] Those skilled in the art can flexibly implement the selection and training of this model according to the inference and capabilities of the above - mentioned model.

[0105] Step S1320: Use a preset graphic - text matching model to determine the graphic - text matching degree between each of the candidate detection result texts and the property operation map corresponding to the project feedback event;

[0106] In one embodiment, the CLIP (Contrastive Language-Image Pre-training) model can be used to efficiently and accurately calculate the text-image matching degree between the candidate detection result text and the property operation map. Specifically: The CLIP model is a multi-modal model that can map text and images into the same vector space and achieve text-image matching by calculating the similarity between vectors. Use the text encoder of the CLIP model to encode each candidate detection result text to generate text embedding vectors. At the same time, use the image encoder of the CLIP model to encode the property operation map corresponding to the project feedback event to generate image embedding vectors. Then, by calculating the cosine similarity between the text embedding vector and the image embedding vector, the text-image matching degree between each candidate detection result text and the property operation map is correspondingly obtained.

[0107] Those skilled in the art can flexibly select and train the model according to the reasoning and capabilities of the above model.

[0108] Step S1330: Determine the detection result text corresponding to the property operation map according to the text-image matching degree, and determine the corresponding progress type text according to whether the quality inspection passes according to the detection result text.

[0109] Select the candidate detection result text with the text-image matching degree exceeding the preset threshold as the detection result text, traverse each detection result text, and determine whether there is a string indicating project completion or a string indicating project incompletion in the detection result text. It can be understood that if the existing string indicates project completion, it means that the quality inspection passes, and project completion is used as the progress type text; if the existing string indicates project incompletion, it means that the quality inspection fails, and project incompletion is used as the progress type text.

[0110] The preset threshold can be set by those skilled in the art as needed.

[0111] Step S1340: Use the optical character recognition algorithm to recognize the characters in the watermark image in the property operation map to obtain the watermark text information, and the watermark information includes the shooting time and the property site location information.

[0112] In accordance with the disclosure of step S1330, this step will not be elaborated here.

[0113] Step S1350: Use a language inference model to generate a progress description text based on the detection result text, the progress type text, and the watermark text information.

[0114] Embed the detection result text, progress type text, and watermark text information into a preset fourth prompt template to obtain a fourth prompt text. Input the fourth prompt text into a language inference model to obtain the project description text output by the model. In one embodiment, the fourth prompt template is: "Please generate a progress description text from the perspective of property management based on the provided detection result text of the property site, the progress type text of the relevant property work, and the watermark text information. The progress description text should clearly and completely describe the operation situation of the property project and the status of the project operation progress. It should include the description of the project operation information and the project operation progress. The provided detection result text: {detected result text to be embedded}, progress type text: {progress type text to be embedded}, watermark text information: {watermark text information to be embedded}."

[0115] To make it clearer for you about the above generation requirements and the generation effects to be achieved, here is an example of a generated progress description text for your reference:

[0116] Demonstration example {Details of the current situation in the example: The project is not completed. The wall repair work has been completed, but the garbage generated from the repair has not been cleaned up yet;

[0117] Progress type text in the example: The project is not completed;

[0118] Watermark text information in the example: Shooting time: Monday, January 15, 2024, 14:30, Property site location information: Corridor on the Xth floor of Building X in XXX Community, Qiwen Road, Haishu District, Ningbo City, Zhejiang Province;

[0119] Progress description text in the example: As of January 15, 2024, the property operation progress of the corridor on the Xth floor of Building X in XXX Community is as follows: The wall repair work has been completed, and the floor cleaning work generated from the repair is in progress. Currently, the overall project progress has not been completed.}

[0120] In this embodiment, through a series of steps of organic integration, an accurate description of the property operation map corresponding to the project feedback event is achieved, and a clear and complete property project progress description text is generated. In this process, the quality inspection experience of property management can be fully utilized to quickly and accurately solve the problems of quality inspection results.

[0121] In a further embodiment, after step S1330, determining the detection result text corresponding to the property operation map according to the graphic-text matching degree, the following steps are included:

[0122] Step S1331: When the detection result text indicates that the quality inspection fails, retrieve each property operation problem in the property management knowledge base that matches the detection result text, obtain the corresponding candidate solutions and their estimated required durations for each property operation problem in the knowledge base, and send them to the competent employee;

[0123] In one embodiment, it is implemented by means of the RAG (Retrieval-Augmented Generation) technology combined with a language inference model. Specifically, first, preprocess the documents in the property management knowledge base, including operations such as text cleaning and chunking. Then, use an embedding model to convert the text into vector representations and store them in a vector database such as Milvus or FAISS. These vector representations can capture the semantic information of the documents and provide a basis for subsequent retrieval. Each document includes a property operation problem and its solution. Those skilled in the art can continuously collect and summarize various property operation problems encountered by competent employees during various property operations, as well as various solutions and their actual required durations for each property operation problem, and flexibly construct the property management knowledge base.

[0124] When the detection result text indicates that the quality inspection fails, use this text as a query input and combine it with the language inference model of RAG. The query text will also be converted into a vector representation and then perform a similarity search in the vector database to find the property operation problem document most relevant to the query text. During the retrieval process, efficient retrieval strategies such as a two-tower model can be adopted to improve the accuracy and efficiency of the retrieval.

[0125] The retrieved relevant documents contain various property operation problems and their corresponding candidate solutions and estimated required durations. By analyzing and integrating these retrieval results through a large language model, a list of candidate solutions for the current detection result text can be generated and sorted according to relevance and feasibility. Send the sorted candidate solutions and their estimated required durations to the competent employee, and the employee can select the most suitable solution according to the actual situation.

[0126] Step S1332: Obtain the solution determined by the competent employee, and append the solution and its estimated required duration to the progress description text.

[0127] Append the solution determined by the competent employee and its estimated required duration to the progress description text so that this part of the content can be appended to the project progress report generated subsequently. Specifically, generating this project progress report can be flexibly implemented by those skilled in the art based on the relevant disclosures in step S1400, and this step will not be elaborated here.

[0128] In this embodiment, first, by means of a language reasoning model combined with RAG, it is possible to quickly and accurately retrieve property operation problems and their candidate solutions that match the detection result text from the property management knowledge base, greatly improving the efficiency of problem location and solution recommendation, reducing the time cost of manual search and analysis, and making full use of the solution experience corresponding to the problems encountered in the property operation process to quickly and accurately solve the corresponding solutions and estimate the processing time. Second, by appending the solutions determined by the competent employees and their estimated required durations to the progress description text, the real-time update and dynamic management of the project progress are realized, enabling project managers to timely understand the latest progress and problem handling situation of the property operation, facilitating the timely adjustment of the project plan and resource allocation. In addition, this combined application also promotes the sharing and inheritance of knowledge. Competent employees can refer to the experience in the knowledge base during the problem-solving process, and at the same time, their solutions will be recorded, enriching the content of the knowledge base and providing valuable references for subsequent property operations. Finally, by generating a project progress report containing the solutions and estimated durations, it is possible to provide more comprehensive and accurate data support for property management decision-making, helping to improve the overall level and service quality of property management.

[0129] Please refer to Figure 2 , a property project work order transfer control device provided to meet one of the purposes of this application, is a functional embodiment of the property project work order transfer control method of this application. The device includes a first event response module 1100, a project work order dispatching module 1200, a second event response module 1300, and a progress report pushing module 1400. Among them, the first event response module 1100 is used to respond to the property project establishment event, call an image processing algorithm to identify the image information of the property current situation map corresponding to the event, and generate a project description text of the property project according to the identification result. The project description text includes the description of the abnormal problems existing in the property site in the map and their processing instructions; the project work order dispatching module 1200 is used to create the project work order of the property project based on the project description text by using a language reasoning model and dispatch the project work order to the appropriate competent employees; the second event response module 1300 is used to respond to the project feedback event of the property project reported by the competent employees, call an image processing algorithm to identify the image information of the property operation map corresponding to the event, and generate a progress description text of the property project according to the identification result. The progress description text includes the description of the project operation information and the project operation progress of the property site in the map; the progress report pushing module 1400 is used to generate a project progress report based on the project work order and its progress description text by using a language reasoning model and push it to a preset interface.

[0130] In a further embodiment, before the project work order dispatching module 1200, it includes: an employee determination sub-module for matching the task description text in the project work order with the function description texts of each employee in the property employee database to determine each candidate suitable employee; an administrator notification sub-module for pushing the project work order associated with each candidate suitable employee and their function description texts to the property administrator; an employee binding sub-module for obtaining the suitable employee determined by the property administrator and binding the suitable employee to the work order.

[0131] In a further embodiment, after the project work order dispatching module 1200, it includes the following steps: an event response sub-module for responding to the event of a suitable employee accepting the order, obtaining the corresponding current captured face image and the project work order, and determining the similarity between the current captured face image and the stored face image of the suitable employee in the property employee database; a work order addition sub-module for adding the project work order to the work order list of the suitable employee when the similarity exceeds a preset threshold.

[0132] In a further embodiment, the second event response module 1300 includes: a text acquisition sub-module for retrieving the project detailed types matching the processing instructions described in the project work order from the quality inspection result database and obtaining the respective candidate detection result texts associated with the project detailed types; a graphic-text matching sub-module for determining the graphic-text matching degree between each of the candidate detection result texts and the property operation diagram corresponding to the project feedback event by using a preset graphic-text matching model; a text determination sub-module for determining the detection result text corresponding to the property operation diagram according to the graphic-text matching degree, and determining the corresponding progress type text according to whether the quality inspection passes according to the detection result text; a first watermark recognition sub-module for performing character recognition on the watermark image in the property operation diagram by using an optical character recognition algorithm to obtain watermark text information, where the watermark information includes the shooting time and the property site location information; a first text generation sub-module for generating a progress description text based on the detection result text, the progress type text, and the watermark text information by using a language inference model.

[0133] In a further embodiment, after the text determination sub-module, it includes: an information sending sub-module for retrieving each property operation problem matching the detection result text from the property management knowledge base when the detection result text indicates that the quality inspection fails, obtaining the candidate solutions corresponding to each property operation problem in the library and their estimated required durations, and sending them to the suitable employee; a text append sub-module for obtaining the solution determined by the suitable employee and appending the solution and its estimated required duration to the progress description text.

[0134] In a further embodiment, the first event response module 1100 includes: a problem determination sub-module, configured to perform object detection on a property current situation map corresponding to the property project establishment event by using a preset target detection model, determine entity status labels corresponding to each object detection target in the map, and determine abnormal problems corresponding to the entity status labels; a project classification sub-module, configured to classify the property current situation map by using a preset project classification model to obtain a project type text; a second optical character recognition sub-module, configured to perform character recognition on a watermark image in the property current situation map by using an optical character recognition algorithm to obtain watermark text information, where the watermark information includes shooting time and property site location information; a second text generation sub-module, configured to generate a corresponding project description text after guiding a language inference model to infer a processing instruction text for the abnormal problem based on the abnormal problem text, the project type text, and the watermark text information.

[0135] In a further embodiment, the second event response module 1300 includes: a text acquisition sub-module, configured to perform object detection on a property operation map corresponding to the project feedback event by using a preset target detection model, determine entity status labels corresponding to all object detection targets in the map, and summarize all entity status labels to obtain a current situation detail text; a progress classification sub-module, configured to classify the property operation map by using a preset progress classification model to determine a corresponding progress type text; a first optical character recognition sub-module, configured to perform character recognition on a watermark image in the property operation map by using an optical character recognition algorithm to obtain watermark text information, where the watermark information includes shooting time and property site location information; a third text generation sub-module, configured to generate a progress description text by using a language inference model based on the current situation detail text, the progress type text, and the watermark text information.

[0136] To solve the above technical problems, an embodiment of the present application further provides a computer device. As Figure 3 shown, it is a schematic internal structure diagram of the computer device. The computer device includes a processor, a computer-readable storage medium, a memory, and a network interface connected through a system bus. Among them, the computer-readable storage medium of the computer device stores an operating system, a database, and computer-readable instructions. The database can store a control information sequence. When the computer-readable instructions are executed by the processor, the processor can implement a property project work order transfer control method. The processor of the computer device is used to provide computing and control capabilities to support the operation of the entire computer device. The memory of the computer device can store computer-readable instructions. When the computer-readable instructions are executed by the processor, the processor can execute the property project work order transfer control method of the present application. The network interface of the computer device is used to connect and communicate with a terminal. Those skilled in the art can understand, Figure 3The structure shown is only a block diagram of some structures related to the solution of this application, and does not constitute a limitation on the computer device to which the solution of this application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine some components, or have a different component layout.

[0137] In this embodiment, the processor is used to execute Figure 2 the specific functions of each module and its sub-modules. The memory stores the program codes and various types of data required to execute the above modules or sub-modules. The network interface is used for data transmission between the user terminal and the server. The memory in this embodiment stores the program codes and data required to execute all modules / sub-modules in the property project work order transfer control device of this application, and the server can call the program codes and data of the server to execute the functions of all sub-modules.

[0138] This application also provides a storage medium storing computer-readable instructions. When the computer-readable instructions are executed by one or more processors, one or more processors are caused to execute the steps of the property project work order transfer control method according to any embodiment of this application.

[0139] Those of ordinary skill in the art can understand that all or part of the processes of implementing the methods in the above embodiments of this application can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above methods. Among them, the aforementioned storage medium can be a computer-readable storage medium such as a magnetic disk, an optical disk, a Read-Only Memory (ROM), or a Random Access Memory (RAM), etc.

[0140] In summary, this application can automatically control the transfer of property project work orders and ensure that the execution process is accurate, reliable, and efficient.

[0141] Those skilled in the art of this technology can understand that the steps, measures, and solutions in the various operations, methods, and processes discussed in this application can be alternated, changed, combined, or deleted. Further, the other steps, measures, and solutions in the various operations, methods, and processes discussed in this application can also be alternated, changed, rearranged, decomposed, combined, or deleted. Further, the steps, measures, and solutions in the prior art that are the same as those disclosed in the various operations, methods, and processes in this application can also be alternated, changed, rearranged, decomposed, combined, or deleted.

[0142] The above are only some embodiments of the present application. It should be noted that for those of ordinary skill in the art, without departing from the principle of the present application, several improvements and refinements can be made, and these improvements and refinements should also be regarded as the protection scope of the present application.

Claims

1. A method for controlling the flow of work orders for a property project, characterized in that: The steps include: In response to a property project establishment event, an image processing algorithm is called to perform image information recognition on the property status map corresponding to the event, and a project description text of the property project is generated according to the recognition result, wherein the project description text includes a description of abnormal problems existing at the property site in the map and instructions for handling the abnormal problems; Using a language reasoning model, creating a project work order for the property project based on the project description text, and assigning the project work order to a suitable employee; In response to the project feedback event reported by the qualified employee for the property project, an image processing algorithm is called to perform image information recognition on the property operation diagram corresponding to the event, and a progress description text of the property project is generated according to the recognition result, wherein the progress description text includes the project operation information of the property site in the diagram and a description of the project operation progress; A language reasoning model is used to generate a project progress report based on the project work order and its progress description text, and push it to a preset interface.

2. The property project work order flow control method according to claim 1 is characterized in that: Before the step of dispatching the project work order to a corresponding qualified employee, the following steps are included: Match the task description text in the project work order with the job description text of each employee in the property employee database to determine each candidate suitable employee; Push the project work order, associated with each candidate qualified employee and their job description text to the property manager; Obtain a qualified employee determined by the property manager, and bind the qualified employee to the work order.

3. The method for controlling the flow of work orders for a property project according to any one of claims 1 to 5, characterized in that: After the step of dispatching the project work order to a suitable qualified employee, the following steps are included: In response to the event of a qualified employee accepting an order, obtain the current facial image and the project work order corresponding to the event, and determine the similarity between the current facial image and the stored facial image of the qualified employee in the property employee database; When the similarity exceeds a preset threshold, the project work order is added to the work order list of the qualified employee.

4. The property project work order flow control method according to claim 1 is characterized in that: The step of calling the image processing algorithm to recognize the image information of the property operation diagram corresponding to the project feedback event and generating a progress description text of the property project according to the recognition result comprises the following steps: Retrieve the project detailed type matching the processing instruction described in the project work order from the quality inspection result library, and obtain each candidate inspection result text associated with the project detailed type; Using a preset image-text matching model to determine the image-text matching degree between each of the candidate detection result texts and the property operation diagram corresponding to the project feedback event; Determine the inspection result text corresponding to the property operation diagram according to the image-text matching degree, determine the corresponding progress type text according to whether the quality inspection is passed according to the inspection result text; Using an optical character recognition algorithm to perform character recognition on the watermark image in the property operation map to obtain watermark text information, wherein the watermark information includes shooting time and property site location information; A language reasoning model is used to generate a progress description text based on the detection result text, progress type text, and watermark text information.

5. The property project work order flow control method according to claim 4 is characterized in that: After the step of determining the detection result text corresponding to the property operation diagram according to the image-text matching degree, the following steps are included: When the test result text indicates that the quality test fails, searching the property management knowledge base for each property operation problem that matches the test result text, obtaining candidate solutions corresponding to each property operation problem in the library and their estimated duration, and sending them to the qualified employee; The solution determined by the qualified employee is obtained, and the solution and its estimated duration are appended to the progress description text.

6. The property project work order flow control method according to claim 1 is characterized in that: The step of calling an image processing algorithm to perform image information recognition on the property status map corresponding to the property project establishment event, and generating a project description text of the property project according to the recognition result, comprises the following steps: Using a preset target detection model to perform target detection on the property status map corresponding to the property project establishment event, determine the entity status label corresponding to each target detection object in the map, and determine the abnormal problem corresponding to the entity status label; Using a preset project classification model to classify the property status map, and obtaining a project type text; Using an optical character recognition algorithm to perform character recognition on the watermark image in the property status map to obtain watermark text information, wherein the watermark information includes shooting time and property site location information; The guided language reasoning model infers the processing instruction text of the abnormal problem based on the abnormal problem text, project type text, and watermark text information, and then generates the corresponding project description text.

7. The property project work order flow control method according to claim 1 is characterized in that: The step of calling the image processing algorithm to recognize the image information of the property operation diagram corresponding to the project feedback event and generating a progress description text of the property project according to the recognition result comprises the following steps: Use a preset target detection model to perform target detection on the property operation map corresponding to the project feedback event, determine the entity status labels corresponding to all target objects in the map, and summarize all entity status labels to obtain status details text; Using a preset progress classification model to classify the property operation diagram, and determining a corresponding progress type text; Using an optical character recognition algorithm to perform character recognition on the watermark image in the property operation map to obtain watermark text information, wherein the watermark information includes shooting time and property site location information; A language reasoning model is used to generate a progress description text based on the current status details text, progress type text, and watermark text information.

8. A property project work order flow control device, characterized in that: include: The first event response module is used to respond to the property project establishment event, call the image processing algorithm to perform image information recognition on the property status map corresponding to the event, and generate a project description text of the property project according to the recognition result, wherein the project description text includes a description of the abnormal problems existing at the property site in the map and the instructions for handling them; A project work order dispatching module is used to create a project work order for the property project based on the project description text by using a language reasoning model, and dispatch the project work order to a suitable employee; The second event response module is used to respond to the project feedback event reported by the qualified employee on the property project, call the image processing algorithm to perform image information recognition on the property operation diagram corresponding to the event, and generate a progress description text of the property project according to the recognition result, wherein the progress description text includes the project operation information of the property site in the diagram and the description of the project operation progress; The progress report push module is used to generate a project progress report based on the project work order and its progress description text by using a language reasoning model, and push it to a preset interface.

9. A computer device comprising a central processing unit and a memory, characterized in that: The central processing unit is used to call and run the computer program stored in the memory to execute the steps of the method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: It stores a computer program implemented according to the method described in any one of claims 1 to 7 in the form of computer-readable instructions, and when the computer program is called and executed by a computer, the steps included in the corresponding method are executed.

Images