Real estate data processing method and device, electronic equipment and storage medium

Abstract

The invention discloses a real estate data processing method and device, electronic equipment and a storage medium, and the method comprises the steps: inputting a real estate right certificate image into a real estate right certificate recognition model, carrying out the semantic segmentation of the real estate right certificate image in a semantic segmentation network, and obtaining a plurality of image regions, geometric information and semantic tags; extracting text blocks in the image area in the text block sorting network, sorting the text blocks according to geometric information to obtain a text block sequence, extracting text contents from the text blocks in the text block sequence in the text recognition network, filling the text contents and semantic tags into a structured data template to obtain structured real estate data, and storing the structured real estate data in the text recognition network. According to the method, the structured real estate data is encrypted and then stored in the database, so that the structured real estate data is obtained by extracting the text content from the real estate certificate through the real estate certificate identification model and filling the text content into the structured data template, manual input is not needed, and the input efficiency and accuracy of the real estate data are improved.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a method, apparatus, electronic device, and storage medium for processing real estate data. Background Technology

[0002] The technology of real estate archive retrieval solutions integrates data science and security technology, and its development is driven by policy promotion, digital transformation and resource management needs.

[0003] In traditional technology, real estate data on property ownership certificates exists in the form of images. In order to transform property ownership certificates into searchable real estate data, the content on the property ownership certificates needs to be manually entered into the database for registration, and then the entered real estate data needs to be manually reviewed and corrected. The level of intelligence and the efficiency of data entry are low. Summary of the Invention

[0004] This invention provides a method, apparatus, electronic device, and storage medium for processing real estate data, in order to solve the problems of low intelligence and low efficiency in manually entering real estate data on real estate ownership certificates.

[0005] In a first aspect, the present invention provides a method for processing real estate data, comprising:

[0006] Obtain the image of the real estate ownership certificate to be entered;

[0007] The real estate ownership certificate image is input into a pre-trained real estate ownership certificate recognition model, which includes a semantic segmentation network, a text block sorting network, and a text recognition network.

[0008] In the semantic segmentation network, the real estate certificate image is semantically segmented to obtain multiple image regions and geometric information and semantic labels for each image region;

[0009] Text blocks in the image region are extracted from the text block sorting network, and the text blocks are sorted according to the geometric information to obtain a text block sequence;

[0010] The text content is extracted from each text block in the text block sequence in the text recognition network.

[0011] The text content and semantic tags are filled into the structured data template to obtain structured real estate data;

[0012] The structured real estate data is encrypted and then stored in a database.

[0013] Secondly, the present invention provides a real estate data processing apparatus, comprising:

[0014] The image acquisition module is used to acquire the image of the real estate ownership certificate to be entered.

[0015] The image input module is used to input the real estate ownership certificate image into a pre-trained real estate ownership certificate recognition model, which includes a semantic segmentation network, a text block sorting network, and a text recognition network.

[0016] The image segmentation module is used to perform semantic segmentation on the real estate certificate image in the semantic segmentation network to obtain multiple image regions and geometric information and semantic labels for each image region.

[0017] A text block sorting module is used to extract text blocks in the image region from the text block sorting network and sort each text block according to the geometric information to obtain a text block sequence.

[0018] The text recognition module is used to extract the text content of each text block in the text block sequence in the text recognition network;

[0019] The structured data generation module is used to fill the text content and the semantic tags into the structured data template to obtain structured real estate data;

[0020] An encrypted storage module is used to encrypt the structured real estate data and then store it in the database.

[0021] Thirdly, the present invention provides an electronic device, the electronic device comprising:

[0022] At least one processor; and

[0023] A memory communicatively connected to the at least one processor; wherein,

[0024] The memory stores a computer program that can be executed by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform the real estate data processing method described in the first aspect of the present invention.

[0025] Fourthly, the present invention provides a computer-readable storage medium storing computer instructions that, when executed by a processor, implement the real estate data processing method described in the first aspect of the present invention.

[0026] In this embodiment of the invention, after acquiring the image of the real estate ownership certificate to be entered, the image is input into a pre-trained real estate ownership certificate recognition model. This model includes a semantic segmentation network, a text block sorting network, and a text recognition network. In the semantic segmentation network, the real estate ownership certificate image is semantically segmented to obtain multiple image regions, as well as geometric information and semantic labels for each region. In the text block sorting network, text blocks are extracted from the image regions, and each text block is sorted according to its geometric information to obtain a text block sequence. In the text recognition network, the text content of each text block in the text block sequence is extracted. The text content and semantic labels are then filled into a structured data template to obtain structured real estate data. The structured real estate data is then encrypted and stored in a database. This invention achieves the goal of segmenting text blocks, sorting text blocks, and extracting text content from the real estate ownership certificate image using the real estate ownership certificate recognition model, and then filling the extracted text content into a structured data template to obtain structured real estate data. This eliminates the need for manual data entry, improving the efficiency and accuracy of real estate data entry.

[0027] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0028] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0029] Figure 1 This is a flowchart of a method for processing real estate data provided in Embodiment 1 of the present invention;

[0030] Figure 2 This is a schematic diagram of the structure of the real estate ownership certificate recognition model;

[0031] Figure 3 This is a schematic diagram of a real estate ownership certificate;

[0032] Figure 4 This is a schematic diagram of the structure of a real estate data processing device provided in Embodiment 2 of the present invention;

[0033] Figure 5 This is a schematic diagram of the structure of the electronic device provided in Embodiment 3 of the present invention. Detailed Implementation

[0034] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0035] Figure 1 This is a flowchart of a method for processing real estate data according to Embodiment 1 of the present invention. This embodiment is applicable to the situation of entering real estate ownership certificates into a database. This method can be executed by a real estate data processing device, which can be implemented in hardware and / or software and can be configured in an electronic device. Figure 1 As shown, the processing method for this real estate data includes:

[0036] S101. Obtain the image of the real estate ownership certificate to be entered.

[0037] In this embodiment, the real estate ownership certificate can be a certificate proving ownership of real estate such as residential buildings and shops, and the relevant information of the real estate ownership certificate needs to be entered into the database.

[0038] In one optional embodiment, an initial image of the real estate ownership certificate to be entered can be acquired by an image acquisition device, and the initial image can be subjected to tilt correction, size adjustment and noise reduction to obtain a real estate ownership certificate image of a preset size.

[0039] Specifically, an image of the real estate ownership certificate to be entered can be captured by a camera or scanner to obtain the initial image of the real estate ownership certificate. The initial image is then further subjected to tilt correction, size adjustment and noise reduction. Tilting can refer to the image being tilted due to improper operation during the scanning or photographing of the real estate ownership certificate. Image tilt will affect the usability of the image, so it is necessary to correct the tilt of the initial image. For example, the real estate ownership certificate image can be adjusted to be horizontal or vertical. Image tilt adjustment can refer to existing technologies, which will not be detailed here.

[0040] Size adjustment can be to adjust the length and width of the initial image to a preset uniform size. Denoising can be any of the existing image denoising techniques. In this embodiment, tilt correction, size adjustment and denoising are performed on the initial image, which can improve the quality of the real estate certificate image and thus improve the accuracy of subsequent text recognition.

[0041] S102. Input the real estate ownership certificate image into the pre-trained real estate ownership certificate recognition model. The real estate ownership certificate recognition model includes a semantic segmentation network, a text block ranking network, and a text recognition network.

[0042] The real estate certificate recognition model in this embodiment can be a CNN (Convolutional Neural Network). This real estate certificate recognition model can include a semantic segmentation network, a text block sorting network, and a text recognition network. The semantic segmentation network is used to perform semantic segmentation on the real estate certificate image to obtain multiple image regions with different semantics. The text block sorting network is used to sort the text blocks in the segmented multiple image regions from top to bottom and from left to right to obtain a text block sequence. The text recognition network is used to recognize the text content of each text block in the text block sequence.

[0043] In one embodiment, the real estate ownership certificate recognition model can be trained through the following steps:

[0044] S1. Construct a real estate certificate recognition model that includes a semantic segmentation network, a text block sorting network, and a text recognition network.

[0045] like Figure 2 The diagram shows the structure of a real estate certificate recognition model, which includes a semantic segmentation network, a text block sorting network, and a text recognition network connected in sequence. Each network can include multiple convolutional pooling layers.

[0046] S2. Obtain a real estate ownership certificate image sample. The real estate ownership certificate image sample is marked with a first image region, the first geometric information of each image region, the first semantic label of each image region, the first text block sequence, and the first text content of the text blocks of each image region.

[0047] like Figure 3 The real estate ownership certificate image sample shown can be labeled with various image areas, such as the real estate ownership number area, the right holder area, the co-ownership status area, and the location area. The first geometric information of each image area can be labeled, which can include the center coordinates, length, and width of each area. The first semantic label can be the field name of each image area, such as "right holder", "co-ownership status", and "location". The first text block sequence can be the reading order of each area in the real estate ownership certificate image sample from top to bottom and from left to right in the image. The first text content can be the content of each area, such as the name of the right holder, the co-ownership status (sole ownership or proportional ownership), and the location (coordinates of the real estate).

[0048] S3. Extract real estate certificate image samples and input them into the real estate certificate recognition model. In the semantic segmentation network, perform semantic segmentation on the real estate certificate image samples to obtain the second geometric information and second semantic labels of multiple second image regions.

[0049] S4. Sort the text blocks in multiple image regions in the text block sorting network to obtain the second text block sequence.

[0050] S5. Extract the text content from the second text block sequence in the text recognition network to obtain the second text content of each text block.

[0051] After the real estate certificate image sample is input into the real estate certificate recognition model, the real estate certificate image sample passes through a semantic segmentation network, a text block sorting network, and a text recognition network to obtain the second geometric information and second semantic label of multiple second image regions, the second text block sequence, and the second text content of each text block in the second text block sequence.

[0052] S6. Calculate the total loss value using the labeled first image region, first geometric information, first semantic label, first text block sequence, first text content, second image region, second geometric information, second semantic label, second text block sequence, and second text content.

[0053] In one embodiment, an image segmentation loss value can be calculated using a first image region, a second image region, first geometric information, and second geometric information, and a semantic segmentation loss value can be calculated using a first semantic label and a second semantic label. The sum of the image segmentation loss value and the semantic segmentation loss value is then calculated to obtain the total segmentation loss value. For example, the similarity can be calculated using the first image region and the second image region, the mean squared error can be calculated using the coordinates, length, and width in the first and second geometric information, the sum of the similarity and the mean squared error can be used as the image segmentation loss value, and the difference between the first semantic label and the second semantic label can be used as the semantic segmentation loss value.

[0054] The ranking loss value is calculated using the first text block sequence and the second text block sequence. For example, the similarity of text blocks at the same position in the first text block sequence and the second text block sequence is calculated, and the average of the similarity at multiple positions is used as the ranking loss value.

[0055] The text extraction loss is calculated using the first and second text contents. For example, the first and second text contents are encoded to obtain encoded features, and the mean squared error of the encoded features is calculated as the text extraction loss.

[0056] In one embodiment, the total loss value can be obtained by summing the total segmentation loss value, the sorting loss value, and the text extraction loss value, or by weighted summing the total segmentation loss value, the sorting loss value, and the text extraction loss value.

[0057] S7. Determine whether the conditions for stopping training are met.

[0058] In this embodiment, the training stop condition can be that the total loss value is less than a preset threshold, or the number of iterations of training reaches a preset number. If the training stop condition is met, then S8 is executed; otherwise, S9 is executed.

[0059] S8. Stop training the real estate ownership certificate recognition model.

[0060] When the conditions for stopping training are met, the real estate certificate recognition model can be determined to have reached the preset accuracy requirements, and training can be stopped to obtain the completed real estate certificate recognition model.

[0061] S9. Adjust the network parameters of the real estate certificate recognition model based on the total loss value, and return to S3.

[0062] In an optional embodiment, the gradient can be calculated based on the total loss value, and the network parameters of the semantic segmentation network, text block sorting network, and text recognition network in the real estate certificate recognition model can be subjected to gradient descent using various gradient algorithms (such as stochastic echelon descent algorithm, batch gradient descent algorithm, etc.) to obtain the real estate certificate recognition model with adjusted network parameters and return to S3 for continued training.

[0063] In another embodiment, a first loss value, a second loss value, and a third loss value can be obtained by calculating the weighted sum of the total loss value, the segmentation loss value, the ranking loss value, and the text extraction loss value. The network parameters of the semantic segmentation network are adjusted based on the first loss value, the network parameters of the text block ranking network are adjusted based on the second loss value, and the network parameters of the text recognition network are adjusted based on the third loss value. This allows the network parameters of each network to be adjusted in a targeted manner according to the function of each network, so that each network can focus more on learning its own function.

[0064] After training, the real estate certificate recognition model in this embodiment can learn the following capabilities: according to the text structure of the real estate certificate, it can segment the real estate certificate image into multiple image regions with semantic labels and geometric information, sort the text blocks of the image regions according to the geometric information to obtain a text block sequence, extract the text content of each text block in the text block sequence, and output the semantic label and its corresponding text content.

[0065] It should be noted that after the real estate certificate recognition model is deployed, it can output only the semantic tags of each text block and its corresponding text content.

[0066] S103. Semantic segmentation of the real estate certificate image is performed in the semantic segmentation network to obtain multiple image regions and geometric information and semantic labels of each image region.

[0067] For example, after the image of a real estate ownership certificate is input into a semantic segmentation network, it can output multiple image regions, as well as the geometric information and semantic labels of each image region, such as... Figure 2 As shown, the image region 1 corresponding to the right holder can be output. The semantic label of the image region 1 is "right holder". The geometric information can be the center coordinates, length and height of the image region 1. The center coordinates can be the coordinates with the upper left corner of the stationary image as the origin.

[0068] S104. Extract text blocks from the image region in the text block sorting network, and sort each text block according to geometric information to obtain a text block sequence.

[0069] In an optional embodiment, the image region includes text blocks. Text blocks can be identified from each image region. Based on the geometric information of the image region, the text blocks identified from the image region are sorted in order from top to bottom and from left to right to obtain a text block sequence. The lower boundary y-coordinate of the text block in the image region is calculated using the center y-coordinate and height of each image region. The right boundary x-coordinate of the text block in the image region is calculated using the center x-coordinate and length of each image region. Each text block is sorted in order of ascending lower boundary y-coordinate. Text blocks with the same lower boundary y-coordinate in the sorted sequence are identified as text blocks in the same row. The text blocks in the same row are adjusted in order of ascending right boundary x-coordinate to obtain the sorted text block sequence.

[0070] Specifically, for each image region, cv2.findContours can be used to determine the text block region, or the smallest bounding rectangle of the black area can be determined after binarizing the image region as the text block region. For each image region, it can be sorted based on the bounding box coordinates x, y, w, h, where x and y are the center coordinates, w is the length, and h is the height. For each text block, the lower boundary y coordinate cy = y + h / 2 is calculated, and then the text blocks are sorted in ascending order of cy to achieve top-to-bottom sorting of the text blocks. Text blocks with the same lower boundary y coordinate cy are in the same row of text blocks. The right boundary x coordinate cx = x + w / 2 can be calculated, and the text blocks in the same row can be sorted according to the right boundary x coordinate cx to obtain the text block sequence sorted from top to bottom and from left to right according to the real estate certificate.

[0071] S105. Extract the text content of each text block in the text block sequence in the text recognition network.

[0072] Each text block is input into the text recognition network, and the output is the text content of that text block. The semantic label of each image region is the semantic label of the text block. That is, for each text block input into the text recognition network, the output is both the semantic label and the text content of the text block. For example, using... Figure 2 Taking image region 1 as an example, the semantic label is "right holder". If the holder of the real estate certificate is "Zhang San", the text block of image region 1 will be input into the text recognition network and the output will be "right holder" & "Zhang San".

[0073] S106. Fill the text content and semantic tags into the structured data template to obtain structured real estate data.

[0074] The structured data template can be a template created according to the reading order of the real estate certificate from top to bottom and from left to right. Each field in the template is blank. After obtaining the semantic tags and text content of each text block, the semantic tags and text content of the text block sequence can be filled into the template in order to obtain JSON structured real estate data.

[0075] S107. Encrypt the structured real estate data and store it in the database.

[0076] In one embodiment, during the transmission of structured real estate data to the database, end-to-end encryption can be implemented at the transport layer using the TLS 1.3 protocol to prevent man-in-the-middle attacks. During data storage, sensitive fields of the structured real estate data (such as the owner's ID number) can be encrypted using AES-256 symmetric encryption. The key is distributed using RSA asymmetric encryption to form a hybrid encryption system. The ciphertext storage supports homomorphic encryption technology, allowing data comparison to be completed in an encrypted state. This is suitable for cross-departmental joint query scenarios. The dynamic key is generated by the server and stored in the HSM hardware module, and is rotated periodically. The client requests data using a pre-set public key.

[0077] In an optional embodiment, when a request for retrieval of structured real estate data is received, it is verified whether the requester has the permission to read the structured real estate data. If so, the permission level of the requester is determined, the structured real estate data is de-identified based on the permission level, and the de-identified structured real estate data is sent to the requester.

[0078] Specifically, the database can provide a structured real estate data API interface to connect with other government departments, enabling them to retrieve real estate data through this API. For example, the permission levels of other government departments can be pre-configured. When a retrieval request for structured real estate data is received from the structured real estate data API interface, the requester's permission level is verified first. Then, the real estate data is retrieved based on the real estate number in the request. Sensitive fields are anonymized according to the permission level before being encrypted and sent to the requester, thereby ensuring the security of the real estate data and protecting the information security of real estate owners.

[0079] In this embodiment of the invention, after acquiring the image of the real estate ownership certificate to be entered, the image is input into a pre-trained real estate ownership certificate recognition model. This model includes a semantic segmentation network, a text block sorting network, and a text recognition network. In the semantic segmentation network, the real estate ownership certificate image is semantically segmented to obtain multiple image regions, as well as geometric information and semantic labels for each region. In the text block sorting network, text blocks are extracted from the image regions, and each text block is sorted according to its geometric information to obtain a text block sequence. In the text recognition network, the text content of each text block in the text block sequence is extracted. The text content and semantic labels are then filled into a structured data template to obtain structured real estate data. The structured real estate data is then encrypted and stored in a database. This invention achieves the goal of segmenting text blocks, sorting text blocks, and extracting text content from the real estate ownership certificate image using the real estate ownership certificate recognition model, and then filling the extracted text content into a structured data template to obtain structured real estate data. This eliminates the need for manual data entry, improving the efficiency and accuracy of real estate data entry.

[0080] Example 2

[0081] Figure 4 This is a schematic diagram of a real estate data processing device provided in Embodiment 2 of the present invention. Figure 4 As shown, the real estate data processing device includes:

[0082] Image acquisition module 401 is used to acquire the image of the real estate ownership certificate to be entered;

[0083] Image input module 402 is used to input the real estate certificate image into a pre-trained real estate certificate recognition model, the real estate certificate recognition model including a semantic segmentation network, a text block sorting network and a text recognition network;

[0084] Image segmentation module 403 is used to perform semantic segmentation on the real estate certificate image in the semantic segmentation network to obtain multiple image regions and geometric information and semantic labels of each image region;

[0085] The text block sorting module 404 is used to extract text blocks in the image region from the text block sorting network and sort each text block according to the geometric information to obtain a text block sequence.

[0086] The text recognition module 405 is used to extract text content for each text block in the text block sequence in the text recognition network;

[0087] The structured data generation module 406 is used to fill the text content and the semantic tags into the structured data template to obtain structured real estate data;

[0088] The encrypted storage module 407 is used to encrypt the structured real estate data and store it in the database.

[0089] Optionally, the image acquisition module 401 includes:

[0090] The image acquisition unit is used to acquire the initial image of the real estate ownership certificate to be entered through the image acquisition device.

[0091] The image preprocessing unit is used to correct, resize, and denoise the initial image to obtain a real estate certificate image of a preset size.

[0092] Optionally, the text block sorting module 404 includes:

[0093] A text block extraction unit is used to identify text blocks from each of the image regions;

[0094] The text block sorting unit is used to sort the text blocks identified from the image region in a top-to-bottom and left-to-right order based on the geometric information of the image region to obtain a text block sequence.

[0095] Optionally, the geometric information includes the center coordinates, length, and height of each image region, and the text block sorting unit is specifically used for:

[0096] The lower boundary y-coordinate of the text block in the image region is calculated using the center y-coordinate of each image region and the height.

[0097] The right boundary x-coordinate of the text block in the image region is calculated using the center x-coordinate of each image region and the length.

[0098] Sort each text block according to the order of its lower boundary y-coordinate from smallest to largest;

[0099] Text blocks with the same lower boundary y-coordinate in the sorting are identified as text blocks on the same line;

[0100] The text blocks in the same row of the sorting process are adjusted according to the order of their right boundary x-coordinates from smallest to largest to obtain the sorted text block sequence.

[0101] Optionally, it also includes a model training module, which is specifically used for:

[0102] A real estate certificate recognition model is constructed, which includes a semantic segmentation network, a text block ranking network, and a text recognition network.

[0103] Obtain a real estate ownership certificate image sample, wherein the real estate ownership certificate image sample is labeled with a first image region, first geometric information of each image region, first semantic label of each image region, first text block sequence, and first text content of text blocks of each image region;

[0104] Extract real estate certificate image samples and input them into the real estate certificate recognition model. Semantic segmentation is performed on the real estate certificate image samples in the semantic segmentation network to obtain the second geometric information and second semantic labels of multiple second image regions.

[0105] The second text block sequence is obtained by sorting text blocks in multiple image regions in the text block sorting network.

[0106] The text content of the second text block sequence is extracted in the text recognition network to obtain the second text content of each text block;

[0107] The total loss value is calculated using the labeled first image region, first geometric information, first semantic label, first text block sequence, first text content, second image region, second geometric information, second semantic label, second text block sequence, and second text content.

[0108] Determine whether the conditions for stopping training are met;

[0109] If so, stop training the real estate certificate recognition model;

[0110] If not, adjust the network parameters of the real estate certificate recognition model based on the total loss value, and return to the step of extracting real estate certificate image samples and inputting them into the real estate certificate recognition model.

[0111] Optionally, the model training module is also specifically used for:

[0112] The image segmentation loss value is calculated using the first image region, the second image region, the first geometric information, and the second geometric information, and the semantic segmentation loss value is calculated using the first semantic label and the second semantic label. The total segmentation loss value is obtained by summing the image segmentation loss value and the semantic segmentation loss value.

[0113] The sorting loss value is calculated using the first text block sequence and the second text block sequence;

[0114] The text extraction loss value is calculated using the first text content and the second text content;

[0115] The total loss value is obtained by summing the segmentation loss value, the ranking loss value, and the text extraction loss value.

[0116] The adjustment of the network parameters of the real estate certificate recognition model based on the total loss value includes:

[0117] The first loss value, the second loss value, and the third loss value are obtained by calculating the weighted sum of the total loss value, the segmentation total loss value, the sorting loss value, and the text extraction loss value, respectively.

[0118] The network parameters of the semantic segmentation network are adjusted based on the first loss value, the network parameters of the text block sorting network are adjusted based on the second loss value, and the network parameters of the text recognition network are adjusted based on the third loss value.

[0119] Optionally, it also includes:

[0120] The retrieval request receiving module is used to verify whether the requester has the permission to read the structured real estate data when it receives a retrieval request from the requester for the structured real estate data.

[0121] The permission determination module is used to determine the permission level of the requesting party;

[0122] The data anonymization module is used to anonymize the structured real estate data based on the permission level.

[0123] The data sending module is used to send the anonymized structured real estate data to the requesting party.

[0124] The real estate data processing apparatus provided in this embodiment of the invention can execute the real estate data processing method provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the method.

[0125] Example 3

[0126] Figure 5A schematic diagram of an electronic device 40 that can be used to implement embodiments of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices (e.g., helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0127] like Figure 5 As shown, the electronic device 40 includes at least one processor 41 and a memory, such as a read-only memory (ROM) 42 or a random access memory (RAM) 43, communicatively connected to the at least one processor 41. The memory stores computer programs executable by the at least one processor. The processor 41 can perform various appropriate actions and processes based on the computer program stored in the ROM 42 or loaded from storage unit 48 into the RAM 43. The RAM 43 may also store various programs and data required for the operation of the electronic device 40. The processor 41, ROM 42, and RAM 43 are interconnected via a bus 44. An input / output (I / O) interface 45 is also connected to the bus 44.

[0128] Multiple components in electronic device 40 are connected to I / O interface 45, including: input unit 46, such as keyboard, mouse, scanner, camera, etc.; output unit 47, such as various types of monitors, speakers, etc.; storage unit 48, such as disk, optical disk, etc.; and communication unit 49, such as network card, modem, wireless transceiver, etc. Communication unit 49 allows electronic device 40 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0129] Processor 41 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 41 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. Processor 41 performs the various methods and processes described above, such as methods for processing real estate data.

[0130] In some embodiments, the method for processing real estate data may be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 48. In some embodiments, part or all of the computer program may be loaded and / or installed on electronic device 40 via ROM 42 and / or communication unit 49. When the computer program is loaded into RAM 43 and executed by processor 41, one or more steps of the real estate data processing method described above may be performed. Alternatively, in other embodiments, processor 41 may be configured to perform the real estate data processing method by any other suitable means (e.g., by means of firmware).

[0131] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations may include: implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0132] Computer programs used to implement the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be performed. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0133] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0134] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0135] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or middleware components (e.g., application servers), or frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0136] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system to address the shortcomings of traditional physical hosts and VPS services, such as high management difficulty and weak business scalability.

[0137] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0138] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A method for processing real estate data, characterized in that, include: Obtain the image of the real estate ownership certificate to be entered; The real estate ownership certificate image is input into a pre-trained real estate ownership certificate recognition model, which includes a semantic segmentation network, a text block sorting network, and a text recognition network. In the semantic segmentation network, the real estate certificate image is semantically segmented to obtain multiple image regions and geometric information and semantic labels for each image region; Text blocks in the image region are extracted from the text block sorting network, and the text blocks are sorted according to the geometric information to obtain a text block sequence; The text content is extracted from each text block in the text block sequence in the text recognition network. The text content and semantic tags are filled into the structured data template to obtain structured real estate data; The structured real estate data is encrypted and then stored in a database.

2. The method according to claim 1, characterized in that, Obtain the image of the real estate ownership certificate to be entered, including: The initial image of the real estate ownership certificate to be entered is captured using an image acquisition device; The initial image is subjected to tilt correction, size adjustment, and noise reduction to obtain a real estate certificate image of a preset size.

3. The method according to claim 1, characterized in that, The text blocks in the image region are extracted from the text block sorting network, and the text blocks are sorted according to the geometric information to obtain a text block sequence, including: Identify text blocks from each of the image regions; Based on the geometric information of the image region, the text blocks identified from the image region are sorted in order from top to bottom and from left to right to obtain a text block sequence.

4. The method according to claim 3, characterized in that, The geometric information includes the center coordinates, length, and height of each image region. Based on the coordinate data of the image regions, the text blocks identified from the image regions are sorted in a top-to-bottom and left-to-right order to obtain a text block sequence, including: The lower boundary y-coordinate of the text block in the image region is calculated using the center y-coordinate of each image region and the height. The right boundary x-coordinate of the text block in the image region is calculated using the center x-coordinate of each image region and the length. Sort each text block according to the order of its lower boundary y-coordinate from smallest to largest; Text blocks with the same lower boundary y-coordinate in the sorting are identified as text blocks on the same line; The text blocks in the same row of the sorting process are adjusted according to the order of their right boundary x-coordinates from smallest to largest to obtain the sorted text block sequence.

5. The method according to claim 1, characterized in that, The real estate ownership certificate recognition model is trained through the following steps: A real estate certificate recognition model is constructed, which includes a semantic segmentation network, a text block ranking network, and a text recognition network. Obtain a real estate ownership certificate image sample, wherein the real estate ownership certificate image sample is labeled with a first image region, first geometric information of each image region, first semantic label of each image region, first text block sequence, and first text content of text blocks of each image region; Extract real estate certificate image samples and input them into the real estate certificate recognition model. Semantic segmentation is performed on the real estate certificate image samples in the semantic segmentation network to obtain the second geometric information and second semantic labels of multiple second image regions. The second text block sequence is obtained by sorting text blocks in multiple image regions in the text block sorting network. The text content of the second text block sequence is extracted in the text recognition network to obtain the second text content of each text block; The total loss value is calculated using the labeled first image region, first geometric information, first semantic label, first text block sequence, first text content, second image region, second geometric information, second semantic label, second text block sequence, and second text content. Determine whether the conditions for stopping training are met; If so, stop training the real estate certificate recognition model; If not, adjust the network parameters of the real estate certificate recognition model based on the total loss value, and return to the step of extracting real estate certificate image samples and inputting them into the real estate certificate recognition model.

6. The method according to claim 5, characterized in that, The total loss value is calculated using the labeled first image region, first geometric information, first semantic label, first text block sequence, first text content, second image region, second geometric information, second semantic label, second text block sequence, and second text content, including: The image segmentation loss value is calculated using the first image region, the second image region, the first geometric information, and the second geometric information, and the semantic segmentation loss value is calculated using the first semantic label and the second semantic label. The total segmentation loss value is obtained by summing the image segmentation loss value and the semantic segmentation loss value. The sorting loss value is calculated using the first text block sequence and the second text block sequence; The text extraction loss value is calculated using the first text content and the second text content; The total loss value is obtained by summing the segmentation loss value, the ranking loss value, and the text extraction loss value. The adjustment of the network parameters of the real estate certificate recognition model based on the total loss value includes: The first loss value, the second loss value, and the third loss value are obtained by calculating the weighted sum of the total loss value, the segmentation total loss value, the sorting loss value, and the text extraction loss value, respectively. The network parameters of the semantic segmentation network are adjusted based on the first loss value, the network parameters of the text block sorting network are adjusted based on the second loss value, and the network parameters of the text recognition network are adjusted based on the third loss value.

7. The method according to any one of claims 1-6, characterized in that, After encrypting and storing the structured real estate data in the database, the process also includes: Upon receiving a retrieval request from a requester for the structured real estate data, verify whether the requester has the permission to read the structured real estate data; If so, determine the permission level of the requesting party; The structured real estate data is anonymized based on the aforementioned permission level; The anonymized structured real estate data is sent to the requesting party.

8. A device for processing real estate data, characterized in that, include: The image acquisition module is used to acquire the image of the real estate ownership certificate to be entered. The image input module is used to input the real estate ownership certificate image into a pre-trained real estate ownership certificate recognition model, which includes a semantic segmentation network, a text block sorting network, and a text recognition network. The image segmentation module is used to perform semantic segmentation on the real estate certificate image in the semantic segmentation network to obtain multiple image regions and geometric information and semantic labels for each image region. A text block sorting module is used to extract text blocks in the image region from the text block sorting network and sort each text block according to the geometric information to obtain a text block sequence. The text recognition module is used to extract the text content of each text block in the text block sequence in the text recognition network; The structured data generation module is used to fill the text content and the semantic tags into the structured data template to obtain structured real estate data; An encrypted storage module is used to encrypt the structured real estate data and then store it in the database.

9. An electronic device, characterized in that, The electronic device includes: At least one processor; and A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform the method for processing real estate data according to any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that, when executed by a processor, implement the method for processing real estate data according to any one of claims 1-7.

Images