| # PDF2Markdown | |
| **Demo:(left: input image; right: rendered markdown.)** | |
|  | |
| 1. Extract PDF features by these tasks: | |
| - Layout Detection: Using the YOLOv8 model for region detection, such as images, tables, titles, text, etc.; | |
| - Formula Detection: Using YOLOv8 for detecting formulas, including inline formulas and isolated formulas; | |
| - Formula Recognition: Using UniMERNet for formula recognition; | |
| - Table Recognition: Using StructEqTable for table recognition; | |
| - Optical Character Recognition: Using PaddleOCR for text recognition; | |
| 2. Convert features to markdown file: | |
| Using simple rules to convert the identified result to markdown (*Note: this is a simply convert code and can only support one-column PDFs, see [MinerU](https://github.com/opendatalab/MinerU) for more complex situation*). | |
| # Usage | |
| ``` | |
| python project/pdf2markdown/scripts/run_project.py --config project/pdf2markdown/configs/pdf2markdown.yaml | |
| ``` | |