21 lines
727 B
Markdown
21 lines
727 B
Markdown
|
# OCR Earsing
|
||
|
|
This repository contains the code for erasing the OCR text from the images. Code is written based on
|
||
|
|
[OCR_SAM]("https://github.com/yeungchenwa/OCR-SAM")
|
||
|
|
|
||
|
|
Have 2 method to earse the text from the image:
|
||
|
|
- Traditional method: using the bounding box of the text to earse the text, and use function OpenCV to earse the text.
|
||
|
|
- Deep learning method: using the bounding box of the text, segment the text based on [SAM]("https://github.com/facebookresearch/segment-anything") (Segment Anything) and use diffuse model to earse the text.
|
||
|
|
|
||
|
|
In my code, I use the deep learning method to earse the text from the image.
|
||
|
|
|
||
|
|
## Installation
|
||
|
|
```shell
|
||
|
|
pip install -r requirements.txt
|
||
|
|
```
|
||
|
|
|
||
|
|
## Run
|
||
|
|
```shell
|
||
|
|
python ocr_eraser.py
|
||
|
|
```
|
||
|
|
|