PDF Processing and Document Analysis API with FastAPI

This application provides an API to process PDF files, extract text, perform OCR, calculate a legibility score, and classify documents using an LLM. Deployed on Heroku - https://ai-powered-doc-classifier-59728d598ae1.herokuapp.com/

Technologies Used

• Python
• FastAPI
• PyMuPDF (fitz)
• Pytesseract
• Pillow (for OCR fallback)
• Ollama (for LLM inference)

Setup and Installation

1. Prerequisites:

   • Python 3.7+
   • Tesseract OCR engine. You can download it from https://github.com/tesseract-ocr/tesseract. Make sure to add the Tesseract executable to your system's PATH.

2. Clone the repository (or download the files):

   git clone <repository_url>
   cd <repository_directory>

3. Create a virtual environment (recommended):

   python -m venv venv
   source venv/bin/activate  # On Windows, use `venv\Scripts\activate`

4. Install the required dependencies:

   pip install -r requirements.txt

5. Download the required LLM models: You need to download the llama3.1:8b-instruct model (and optionally mistral:7b-instruct for verification) for Ollama:

   ollama pull llama3.1:8b-instruct
   ollama pull mistral:7b-instruct

Open your browser and navigate to http://127.0.0.1:8001/docs. You will see the Swagger UI, which allows you to interact with the API.

API Endpoint

POST /classify/

• Description: Upload a PDF file to be processed and classified.
• Request:
  • file: The PDF file to be uploaded.
• Response: A JSON object containing the processing results, including:
  • filename: The name of the uploaded file.
  • page_count: The total number of pages in the PDF.
  • average_legibility_score: The average legibility score across all pages.
  • total_images: The total number of images found in the PDF.
  • pages: A list of results for each page, including extracted text, OCR text, legibility score, image count, and bounding boxes for found matches.
  • document_classification: The LLM-based classification result for the entire document.