Extract PDF Table to JSON with ComPDFKit

By ComPDFKit | 2023 Dec 01
Conversion SDK Data Extraction

Despite the widespread use of PDFs, the need to extract data from PDFs is particularly prevalent in businesses. The characteristics of PDF formats make their data challenging to access or obtain. So, some companies, such as ComPDFKit, specialize in extracting and utilizing PDF data.


Let's delve into the considerations crucial for converting tabular data from PDFs to JSON format. By the end, you'll have the opportunity to enjoy ComPDFKit API's free monthly 1000 file conversion instantly.



Why Extracting Tables into JSON Data?


Table extraction is a technology for recognizing and extracting table data from PDF files, making data more readable, processable, and programmable language interactive. Converting PDF into JSON, the structured data, for reuse within any desired business process, greatly aids the automation of enterprise workflows. There are numerous scenarios for table extraction, including:


         - Robotic Process Automation (RPA): Harnessing cutting-edge AI and deep learning algorithms to automate office OA and factory OT workflows — boosting efficiency and reducing costs. This enables monitoring and reporting as well as complex decision-making and self-optimization.

         - PDF Data Extraction and Intelligent Analysis: Leveraging advanced OCR and natural language processing to extract PDF data, facilitating high-level AI analysis. Providing a user-friendly interface and robust API, even non-technical staff can effortlessly perform data extraction and analysis.

         - In finance, accounting, auditing, and sales industries, table extraction can automatically fill financial report data, saving time and effort, and preventing manual inaccuracies.

         - Within education, research, and healthcare, extracted table data can be used for statistics and visualizations, enhancing the readability and credibility of documents like papers, reports, and medical records.

         - In legal, government, and commercial sectors, ensure data accuracy and security by using table extraction for verification from documents like contracts, invoices, and application forms.


In summary, table extraction is an efficient data processing method, enabling valuable information retrieval from PDF files to optimize workflows and decision-making processes.

Differences Between PDF Data and JSON Data


PDF and JSON are fundamentally different file formats for data storage and presentation. This section delineates these differences and their respective advantages and disadvantages.


PDF Data

PDF (Portable Document Format), a graphical file format at its core, ensures layout fidelity, supporting text, fonts, vector graphics, raster images, and the information required for display. Thanks to its vector graphics preservation, PDFs offer high print quality.


PDF data is static and unamenable to dynamic modifications or analysis. It's not human-readable without specific software tools and is unstructured, often presented as text, images, and tables, which makes direct data access difficult.



JSON, a file format for storing data, is an open standard file and data interchange format, using human-readable text to store and transmit data objects consisting of attribute-value pairs and arrays (or other serializable values). It's a popular data format with varied uses in data exchange. Supported data types include numbers, strings, booleans, arrays, objects, and null.


Data in JSON is dynamic, structured, lightweight, modifiable, analyzable, and easily expresses complex data structures in concise text, facilitating interaction with various programming languages.

Challenges in Extracting PDF Data


There are two major hurdles when extracting data from PDFs: the diversity and complexity of table types and text within tables.


Firstly, tables are not always uniform—irregular tables, borderless tables, half-bordered tables, tables spanning multiple pages, and even those within images complicate data extraction. These irregularities prevent improvements in extraction precision by using algorithmic, but ComPDFKit's Document AI technology significantly enhances outcomes.


Secondly, the challenge of varied text styles within tables, different languages, text within images, and invisible text nested within PDF documents must be considered. OCR technology solves image-based text issues, and ComPDFKit could recognize an extensive variety of language types.


Though extracting data from PDFs has its difficulties, technological advancements have paved the way for effective tools and methodologies to help us capture and organize this data. ComPDFKit continues to explore and enhance our technology, striving to deliver the optimal data extraction solutions for your projects.

PDF Table Extraction Results


Using the same 1000 sample data sets (50% standard tables and 50% non-standard), ComPDFKit Document AI's Table Recognition Versions 1 and 2, as well as well-known PDF technologies company H and P, were tested for table recognition accuracy. The comparative table includes the following three key metrics:


         - Number of rows and columns in a table.

         - Cell information (number of cells, accuracy of merged cells).

         - Table content (accuracy of cell content).


A successful table recognition is counted only when all three indicators are met. Here are the specific test results:

PDF Table Extraction Results


The data from these tests indicate that the ComPDFKit Document AI Table Recognition Version 2 has improved the accuracy rate for standard table extraction by 20% compared to Version 1. Also, it outperforms other brands in table extraction effectiveness. Brand P shows the best performance for non-standard table recognition, but overall, Document AI V2 version has the most robust results.


1. Example of Standard Table Recognitions:

         - Original Image: 

Example of Standard Table Recognitions


         - Recognition Result:

Recognition Result


2. Example of Non-Standard Table Recognitions:


  • Original Image:

Example of Non-Standard Table Recognitions


  • Recognition Result:

Recognition Result



Customize Your PDF to JSON Data Extraction


The PDF to JSON functionality of ComPDFKit API supports, not only table extraction but also the retrieval of text, images, and more from PDF files. 


         - Enable AI Table Recognition: Utilize ComPDFKit Document AI technology for table recognition.

         - Extract Text Only: Retrieve all text content within a PDF, including text in images if OCR is enabled.

         - Extract PDF Images: This function is not compatible with OCR.

         - Extract Tables Only: Retrieve the structure and internal data of tables.

         - Enable OCR.

         - Choose OCR Language.


Tailor your PDF data extraction according to project needs, and output extracted data in various file types such as JSON, Word, Excel, PPT, HTML, CSV, RTF, TXT, etc.

Free Trial Tool!


Start experiencing ComPDFKit's data extraction functionalities now!

        - Try ComPDFKit Free Online Tool.

        - ComPDFKit API calls, with 1000 free API requests per month.