Skip to main content

Table Outliner Textract

Overview​

Table Outliner Textract is a model designed to detect and extract tables from document images using the AWS Textract service. The model processes each page of a document, leveraging AZURE OCR and table detection capabilities to accurately identify table boundaries, rows, columns, and cells. It then structures the extracted information into a format that can be easily processed for further analysis.

How It Works​

  1. Image Preprocessing: For each page of the document:

    • The image is prepared for processing, ensuring it meets the input requirements of AWS Textract.
    • Basic image preprocessing steps such as resizing, binarization, and noise reduction may be applied to enhance detection accuracy.

  2. Textract Integration: The processed images are sent to AWS Textract, which:

    • Detects and extracts text from tables within the document.
    • Identifies the structure of tables, including rows, columns, and cells.

  3. Line and Table Structure Identification: AWS Textract returns detailed information about the position and content of detected tables, which includes:

    • Coordinates of table boundaries.
    • Locations of individual cells, rows, and columns.

  4. Data Extraction and Post-Processing: The extracted data is further processed by:

    • Converting the positional data into a structured format.
    • Organizing the detected tables according to their spatial arrangement on the page.

  5. Validation and Filtering: The model ensures that:

    • Only valid table structures are retained.
    • Any extraneous or misidentified lines are filtered out to improve the accuracy of the table extraction.

  6. Final Output: The tables, along with their contents, are formatted into a structured text format, where each table is represented in json format, making them ready for further analysis or export to other applications.

Installation

Prerequisites​

Ensure you have the following before starting the installation:

  • Git installed on your machine
  • Python and pip installed
  • FURY authentication token

Installation Steps​

  1. Clone the repository

    git clone https://github.com/cognaize/ai-core-table-outliner-textract/tree/prod

  2. Navigate to the project directory

    cd ai-core-table-textract

  3. Install the required Python packages

    FURY_AUTH=${FURY_AUTH} pip install -r requirements.txt

Usage​

Here's a quick start example to use table outliner textract:

# Create an instance of TableTextFormatter with the specified PDF path
model = TableTextFormatter(pdf_path='/path_to_your_document.pdf')

# Predict tables into strucutred text
tables_str = model.to_structured_text_table()

# Print the tables
print(tables_str)