gpt4ocr
| Crates.io | gpt4ocr |
| lib.rs | gpt4ocr |
| version | 0.3.2 |
| created_at | 2024-08-06 08:40:45.098375+00 |
| updated_at | 2024-08-06 08:55:39.240404+00 |
| description | Extract structured text from PDFs using OpenAI's GPT4o. |
| homepage | |
| repository | |
| max_upload_size | |
| id | 1327046 |
| size | 4,953,123 |
Dhruvajyoti Sarma (dhruva71)
documentation
README
GPT4OCR
A simple OCR tool that uses GPT-4o to perform OCR on pdf files. Requires a .env file with the following variables:
OPENAI_API_KEY=your_openai_api_key
Alternatively, you can pass in the OPENAI_API_KEY as an environment variable to the extract_json_from_pdf function.
Operating systems
Runs on linux. Needs poppler-utils to be installed. To install it on Ubuntu, run
sudo apt install poppler-utils
sudo apt install libssl-dev
Important observations
- The time grows with the number of fields generated. You can specify the JSON format to limit the number of fields generated in the prompt, and that can help reduce the time required.
- JSON comes back as a markdown block, so you can remove the "
json" and "" to get the JSON data. This handled by the library currently.
Pending
- Parallel processing to speed up the process.