Table generation method and system based on AI auxiliary annotation and webpage rendering

Abstract

The invention discloses a table generation method and system based on AI auxiliary annotation and webpage rendering. The method comprises the steps that AI auxiliary annotation is carried out; synthesizing a webpage rendering table; and exporting data and generating a training sample. The generation efficiency and quality of the table recognition model training data are improved in an automatic mode, and the problems that in the prior art, manual labeling is high in cost, low in efficiency, prone to making mistakes and the like are solved. The AI technology is used for assisting in table structure labeling, high-quality and diversified table data are generated in combination with the webpage rendering technology, and the training requirement of a table recognition model can be met.

Description

Technical Field

[0001] The present invention belongs to the technical field of table generation, and specifically relates to a table generation method and system based on AI-assisted annotation and web page rendering. Background Art

[0002] In recent years, table recognition technology has been widely used in fields such as document processing and data mining. However, training a high-precision table recognition model relies on a large number of well-annotated training samples, and the acquisition and annotation of these samples are often time-consuming and laborious, especially in tables with complex formats. Traditional manual annotation methods not only have high costs but are also prone to errors, making it difficult to ensure the consistency and accuracy of annotations. To address this challenge, AI-assisted annotation has gradually become an effective solution. Through AI technology, it is possible to automatically detect and preliminarily annotate the structure and content of tables, significantly reducing the workload and error rate of manual annotation.

[0003] On this basis, further using web page rendering technology for table synthesis can generate a high-quality data set similar to the tables in actual documents. This synthesis method can not only highly flexibly define the table style and content but also ensure the diversity and authenticity of the generated data, providing rich sample resources for the training of table recognition models. Therefore, combining AI-assisted annotation and web page rendering technology has important application value in providing an efficient and automated data generation approach for table recognition models.

[0004] Existing technical solutions adopt the method of manual annotation. By formulating standard annotation rules (quadrilateral annotation for each cell frame of the table and identifying the text information within the cell), a large number of outsourced personnel are recruited for annotation or directly purchase the annotation services of annotation companies.

[0005] The existing technology has the following four disadvantages:

[0006] 1. The annotation process is time-consuming and laborious: Manually annotating a large number of samples requires a large amount of time and manpower. Especially when facing complex table formats, the annotation difficulty increases.

[0007] 2. Prone to errors and difficult to ensure consistency: Due to the diverse table structures, the manual annotation process is prone to errors, and the quality of the annotation results is difficult to ensure consistency.

[0008] 3. High cost: The high labor intensity and long-time investment in manual annotation result in a relatively high overall annotation cost, especially obvious in the case of large-scale data sets.

[0009] 4. Difficult to meet the demand for high-quality data: The existing manual annotation method is less efficient in generating high-quality and diverse training data and is difficult to quickly respond to the model training requirements. Summary of the Invention

[0010] In view of the above problems, the present invention is proposed to provide a table generation method and system based on AI-assisted annotation and web page rendering that overcome the above problems or at least partially solve the above problems.

[0011] To achieve the above object, the present invention adopts the following technical solutions:

[0012] A table generation method based on AI-assisted annotation and web page rendering, the method includes:

[0013] AI-assisted annotation;

[0014] Web page rendering table synthesis;

[0015] Data export and training sample generation.

[0016] Optionally, the AI-assisted annotation includes:

[0017] Table detection, detecting the table area in the document through an AI algorithm to locate the boundary and structure information of the table;

[0018] Structure annotation, after table detection, the AI algorithm automatically annotates the row and column structure of the table, identifies the text, numbers and symbols in the table, and generates preliminary structured data.

[0019] Optionally, the boundary and structure information of the table includes row, column and cell positions.

[0020] Optionally, the web page rendering table synthesis includes:

[0021] Table style definition, defining the table style through web page rendering technology to ensure that the generated table is visually consistent with the real table;

[0022] Content filling and data generation, using the AI annotation results to fill the structured data of the table into the table template defined by the web page rendering to generate diverse table sample data.

[0023] Optionally, the table style includes the format, border, font and color of the table.

[0024] Optionally, the data export and training sample generation includes:

[0025] Sample export, exporting the tables generated through web page rendering in different formats to meet the diverse data input requirements for model training;

[0026] Training sample generation, generating high-quality table data that meets the training requirements of the table recognition model for use by the model.

[0027] The present invention also provides a table generation system based on AI-assisted annotation and web page rendering, which is used to execute a table generation method based on AI-assisted annotation and web page rendering described in any one of the foregoing, including:

[0028] A table detection module, which is used to detect the table area in a document and identify the boundaries of the table;

[0029] A table structure recognition module, which is used to automatically recognize the row, column, and cell structures of a table and extract the table content;

[0030] A rendering module, which is used to convert the annotated table structure into web page code and generate the visual style of the table according to user settings;

[0031] A style customization module, which is used to allow users to customize the table style through an interface to ensure that the generated table sample meets the actual requirements;

[0032] A format conversion module, which is used to export the rendered table into multiple formats for use in training a table recognition model.

[0033] In summary, due to the adoption of the above technical solutions, the beneficial effects of the present invention are as follows:

[0034] 1. The present invention improves the generation efficiency and quality of training data for a table recognition model through an automated method, overcoming the problems of high manual annotation costs, low efficiency, and easy errors in the prior art. By using AI technology to assist in table structure annotation and combining web page rendering technology to generate high-quality and diverse table data, the training requirements of the table recognition model can be met.

[0035] 2. The present invention automatically detects the table area through an AI algorithm and annotates the table structure and content, realizing the automated annotation of the table structure and reducing the workload and error rate of manual annotation; the present invention supports exporting the generated table in multiple formats (such as pictures, HTML, Excel, etc.) to meet the training requirements of different table recognition models.

[0036] 3. The present invention combines AI-assisted annotation with web page rendering technology to form an efficient and automated table generation method; through the style customization function in the web page rendering module, users can flexibly adjust the visual style and layout of the generated table to make it more consistent with the actual table. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 It is a schematic flowchart of a table generation method based on AI-assisted annotation and web page rendering provided by an embodiment of the present application;

[0038] Figure 2Schematic diagram of a table generation system based on AI-assisted annotation and web page rendering provided by an embodiment of the present application. Detailed implementation manners

[0039] To make the objectives, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0040] Embodiment 1:

[0041] Please refer to Figure 1 , this embodiment provides a table generation method based on AI-assisted annotation and web page rendering, and the method includes:

[0042] S1. AI-assisted annotation.

[0043] AI-assisted annotation includes:

[0044] Table detection, detecting the table area in the document through an AI algorithm to locate the boundary and structure information of the table; the boundary and structure information of the table includes row, column and cell positions.

[0045] Structure annotation, after table detection, the AI algorithm automatically annotates the row and column structure of the table, identifies the content such as text, numbers and symbols in the table, and generates preliminary structured data.

[0046] S2. Web page rendering table synthesis.

[0047] Web page rendering table synthesis includes:

[0048] Table style definition, defining the table style through web page rendering technology to ensure that the generated table is visually consistent with the real table, and the table style includes the format, border, font and color of the table.

[0049] Content filling and data generation, using the AI annotation results to fill the structured data of the table into the table template defined by web page rendering to generate diverse table sample data.

[0050] S3. Data export and training sample generation.

[0051] Data export and training sample generation includes:

[0052] Sample export: Export the table generated by web page rendering in different formats, such as pictures, HTML, Excel, etc., to meet the diverse data input requirements of model training.

[0053] Training sample generation: Generate high-quality table data that meets the training requirements of the table recognition model for use by the model.

[0054] This embodiment improves the generation efficiency and quality of the training data of the table recognition model in an automated manner, overcoming the problems of high cost, low efficiency, and easy errors in manual annotation in the prior art. By using AI technology to assist in table structure annotation and combining web page rendering technology to generate high-quality and diverse table data, it can meet the training requirements of the table recognition model.

[0055] This embodiment automatically detects the table area through an AI algorithm and annotates the table structure and content, realizing the automated annotation of the table structure and reducing the workload and error rate of manual annotation; this embodiment supports exporting the generated table in multiple formats (such as pictures, HTML, Excel, etc.) to meet the training requirements of different table recognition models.

[0056] This embodiment combines AI-assisted annotation with web page rendering technology to form an efficient and automated table generation method; through the style customization function in the web page rendering module, users can flexibly adjust the visual style and layout of the generated table to make it more consistent with the actual table.

[0057] Embodiment Two:

[0058] Please refer to Figure 2 , Embodiment Two discloses a table generation system based on AI-assisted annotation and web page rendering, which is used to execute the table generation method based on AI-assisted annotation and web page rendering described in Embodiment One, including:

[0059] Table detection module, used to detect the table area in the document and identify the boundaries of the table.

[0060] Table structure recognition module, used to automatically identify the row, column, and cell structures of the table and extract the table content.

[0061] Rendering module, used to convert the annotated table structure into web page code and generate the visual style of the table according to user settings.

[0062] Style customization module, used to allow users to customize the table style through the interface to ensure that the generated table samples meet the actual requirements.

[0063] Format conversion module, used to export the table generated by rendering into multiple formats for use in training the table recognition model.

[0064] The above content is a further detailed description of the present invention in combination with specific preferred embodiments, and it cannot be determined that the specific implementation of the present invention is only limited to these descriptions. For those of ordinary skill in the technical field to which the present invention pertains, without departing from the concept of the present invention, several simple deductions or substitutions can still be made, and all should be regarded as belonging to the protection scope of the present invention.

Claims

1. A table generation method based on AI-assisted annotation and web page rendering, characterized in that: The method comprises: AI-assisted labeling; Web page rendering table synthesis; Data export and training sample generation.

2. A table generation method based on AI-assisted annotation and web page rendering as claimed in claim 1, characterized in that: AI-assisted annotation includes: Table detection: Use AI algorithms to detect table areas in documents and locate table boundaries and structural information; Structural annotation: After table detection, the AI ​​algorithm automatically annotates the row and column structure of the table, identifies the text, numbers and symbols in the table, and generates preliminary structured data.

3. A table generation method based on AI-assisted annotation and web page rendering as claimed in claim 2, characterized in that: The table's boundary and structure information includes row, column, and cell locations.

4. The table generation method based on AI-assisted annotation and web page rendering according to claim 1, characterized in that: Web page rendering table synthesis includes: Table style definition: Define the table style through web page rendering technology to ensure that the generated table is visually consistent with the real table; Content filling and data generation, using AI annotation results, fill the structured data of the table into the table template defined by web page rendering to generate diverse table sample data.

5. A table generation method based on AI-assisted annotation and web page rendering as claimed in claim 4, characterized in that: The table style includes the format, border, font and color of the table.

6. The table generation method based on AI-assisted annotation and web page rendering according to claim 1, characterized in that: Data export and training sample generation include: Sample export: export the tables generated by web page rendering in different formats to meet the diverse data input requirements of model training; Training sample generation: Generate high-quality table data that meets the training requirements of the table recognition model for use by the model.

7. A table generation system based on AI-assisted annotation and web page rendering, which is used to execute a table generation method based on AI-assisted annotation and web page rendering according to any one of claims 1 to 6, characterized in that: include: A table detection module is used to detect the table area in the document and identify the boundaries of the table; Table structure recognition module, used to automatically identify the row, column, and cell structure of a table and extract the table content; The rendering module is used to convert the marked table structure into web page code and generate the visual style of the table according to the user settings; The style customization module is used to allow users to customize the table style through the interface to ensure that the generated table sample meets actual needs; The format conversion module is used to export the rendered tables into multiple formats for use in table recognition model training.

Images