Use SAM2 to extract and warp a page from a photo.
Project description
Page Extractor
Extracts a page from a photo, and warps it to a rectangular image using skimage.
Features
- Remove margin clutter from photos of pages, usually improved document processing.
- GroundingDINO detection model integration.
- SAM 2.1
- Customizable text prompt, e.g. to "receipt." or "invoice."
Getting Started
Prerequisites
- Python 3.10 or higher
Installation
Installing PyTorch Dependencies
Before installing pageextractor, install PyTorch:
pip install torch==2.4.1 torchvision==0.19.1 --extra-index-url https://download.pytorch.org/whl/cu124
Installation options
pip install -U git+https://github.com/UG-Team-Data-Science/pageextractor.git
Or:
git clone https://github.com/UG-Team-Data-Science/pageextractor && cd pageextractor
pip install -e .
Usage
from PIL import Image
from matplotlib import pyplot as plt
from pageextractor import PageExtractor
img = Image.open('example.png')
model = PageExtractor(sam_type='sam2.1_hiera_tiny', device='cuda')
mask, polygon, cropped = model.extract_page(img)
_, (ax0, ax1, ax2) = plt.subplots(1, 3, figsize=(30, 15))
ax0.imshow(img)
ax1.imshow(img)
ax1.plot(*polygon[[0,1,2,3,0]].T, 'r:')
ax1.imshow(1-mask, cmap='Blues', alpha=0.8 - 0.8*mask)
ax2.imshow(cropped)
Examples
Photo of a book page, unfortunately centerfold and page block are included
Photo of a letter
Acknowledgments
This project is based on/used the following repositories:
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Filter files by name, interpreter, ABI, and platform.
If you're not sure about the file name format, learn more about wheel file names.
Copy a direct link to the current filters
File details
Details for the file pageextractor-0.1.7.tar.gz.
File metadata
- Download URL: pageextractor-0.1.7.tar.gz
- Upload date:
- Size: 10.1 kB
- Tags: Source
- Uploaded using Trusted Publishing? No
- Uploaded via: uv/0.7.13
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
902ebe3dc9cb7046d3e831f3ee4d1819a9f3f602369d18f31d297acaca2c0b31
|
|
| MD5 |
4855d4bdb5699d9d9caee7ff3f919a42
|
|
| BLAKE2b-256 |
74390e18769849d359543fae178d02a127eb7e7e89a14b33f6e0bdec8002bb36
|
File details
Details for the file pageextractor-0.1.7-py3-none-any.whl.
File metadata
- Download URL: pageextractor-0.1.7-py3-none-any.whl
- Upload date:
- Size: 10.7 kB
- Tags: Python 3
- Uploaded using Trusted Publishing? No
- Uploaded via: uv/0.7.13
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
45076118eaef527febe24e43a730f2e9b51a8c5581713e11b37dac33a4671190
|
|
| MD5 |
d8791f4a12eac8564a2c165126e374bc
|
|
| BLAKE2b-256 |
d8db41fb0d4c80379ba09add4e8d2e325fe1a08b6238283f11e4a827d83b1fc8
|