Image Text Extraction Using OpenAI GPT-4

This Python script extracts text from images using OpenAI's GPT-4 API. The script supports both image URLs and base64-encoded images as inputs. Extracted text can be saved to a local file for further use.

---

Features

• Text Extraction from Image URLs: Extract text directly from an image available online.
• Text Extraction from Base64 Images: Extract text from locally stored images by converting them into base64 encoding.
• Accurate Text Parsing: Ensures completeness by carefully extracting all text elements, including faint or small text, and preserving the original structure.
• Output to File: Saves extracted text to a specified file.

---

Requirements

• Python 3.7+
• OpenAI Python SDK
• Base64 library (standard library)

---

Installation

1. Clone this repository or download the script.
2. Install the required dependencies:

   pip install openai

3. Ensure you have an OpenAI API key.

---

Usage

Extracting Text from an Image URL

1. Uncomment the relevant section in the script:

   # image_url = "https://example.com/image.jpg"
   # extracted_text = image_to_text_from_url(image_url)
   # output_file_path = "path/to/extracted_text.txt"
   # with open(output_file_path, "a", encoding="utf-8") as text_file:
   #     text_file.write(extracted_text)
   #     text_file.write("\n")

2. Replace the image_url with the URL of your image.
3. Run the script and view the extracted text in the specified file.

Extracting Text from a Local Image

1. Provide the path to your local image:

   local_image_path = "path/to/image.png"

2. The script will automatically convert the image to base64 encoding and extract text.
3. View the extracted text in the console or in the specified output file.

---

Code Structure

Functions

1. image_to_base64(image_path)

   • Converts a local image to base64 encoding.

2. image_to_text_from_url(image_url)

   • Extracts text from an image using its URL.

3. image_to_text_from_base64(image_base64)

   • Extracts text from an image provided in base64 format.

Example Workflow

• Input: Image URL or local image file.
• Process:
  • For URLs: Direct API call.
  • For local files: Convert to base64, then API call.
• Output: Extracted text saved to a file.

---

Configuration

• Update your OpenAI API key in the client initialization:

  client = OpenAI(
      api_key="your-openai-api-key"
  )

• Specify the file path for saving the extracted text.

---

Limitations

• The script requires an active OpenAI API key with appropriate permissions.
• Limited by the capabilities of the GPT-4 model for OCR tasks.

---

Contributing

Feel free to fork this repository and submit pull requests for new features or improvements.

---

License

This project is licensed under the MIT License. See the LICENSE file for details.