drakonis96/ocrai repository overview

⁠ocrAI 🤖

🆕 Enhanced OCR + AI Mode: The AI-corrected text is now embedded directly into the PDF, preserving the original layout and positioning within the document.

📄 New Output Format in Full AI OCR Mode: The output is now a Markdown-formatted .txt file, structured with titles, subtitles, and page markers. This file can be converted directly within the app to a clean, paginated PDF, making it ideal for tasks like translating books while preserving formatting and pagination.

ocrAI is a simple web app that combines Optical Character Recognition (OCR) and Artificial Intelligence (AI) to process and translate documents. It offers an intuitive interface with real-time feedback and emoji progress.

⁠Key Features

• File Management 📤

  • Upload PDF or image files easily.
  • Files are saved with unique names to avoid overwrites.
  • Delete all files with one click (removes from both uploads and outputs).

• OCR Processing Modes 🔍

  • OCR (Tesseract Only):
    Extracts text with Tesseract and embeds it into the PDF using OCRmyPDF. The TXT file contains the raw OCR output.
  • OCR + AI (Tesseract + AI):
    Uses Tesseract to extract text, then sends it to Gemini AI for correction and formatting. The corrected text is embedded directly into the resulting PDF, and a TXT file is also generated.
  • AI (Full AI OCR):
    Uses Gemini AI to process the document page by page. Generates a TXT file in markdown format (with page markers and structure). This TXT can be converted to a PDF using the TXT to PDF tool in the app.
  • All modes show real-time progress with emojis and run in the background.

• TXT to PDF 📝

  • Convert any TXT file (especially markdown from AI mode) to a clean, paginated PDF directly from the app.

• Translation 🌐

  • Translate PDF or TXT documents page by page.
  • Progress updates are displayed, and a TXT file with the final translation (including page markers) is generated.

• Configuration ⚙️

  • Manage and add new Gemini AI models and languages.
  • Update or add custom prompts for OCR, correction, and translation.
  • Download or upload the complete configuration (prompts and models).

⁠How to Use the Application

1. Upload and Process Files:

   • Go to the OCR tab.
   • Select your file (PDF or image).
   • Choose one of the processing modes:
     • OCR (Tesseract Only)
     • OCR + AI (Tesseract + AI for correction and AI-embedded PDF)
     • AI (Full AI OCR, generates markdown TXT)
   • For AI modes, select the desired prompt.
   • Click Upload and process and watch the real-time progress.
   • If you used AI mode, you can convert the resulting TXT to PDF using the TXT to PDF tab.

2. Translate Documents:

   • Go to the Translation tab.
   • Upload a new file or select one from the list of processed files.
   • Choose the target language and translation prompt.
   • Click Translate and observe the progress as each page is processed.
   • The result is saved in a TXT file with page markers.

3. View Processed Files:

   • Go to the Processed Files tab.
   • Download or delete files (with confirmation prompts).

   Note: The Processed Files list is ordered from most recent to oldest (top = newest).

4. Configure the Application:

   • Go to the Configurations tab.
   • Add, edit, or delete custom prompts.
   • Manage Gemini models: add new models or delete existing ones.
   • Configure languages and download or upload the complete configuration.

5. Convert TXT to PDF:

   • Go to the TXT to PDF tab.
   • Select a TXT file (for example, one generated by AI mode).
   • Click Convert to PDF to get a clean, paginated PDF.

⁠How to Run ocrAI

⁠Prerequisites

• Docker
• Docker Compose

⁠Build and Run

docker-compose up --build

Then, open your browser at http://localhost:5015⁠ to start using ocrAI.

Technologies Used
Frontend: React, Axios
Backend: Flask, Python
OCR: Tesseract, pdf2image, OCRmyPDF
AI: Gemini API
Containerization: Docker, Docker Compose