/P2PaLA

Page to PAGE Layout Analysis Tool

Primary LanguagePythonGNU General Public License v3.0GPL-3.0

P2PaLA

❗❗ P2PaLA is deprecated ❗❗

Python Version Code Style

Page to PAGE Layout Analysis (P2PaLA) is a toolkit for Document Layout Analysis based on Neural Networks.

💥 Try our new DEMO for online baseline detection. ❗❗

If you find this toolkit useful in your research, please cite:

@misc{p2pala2017,
  author = {Lorenzo Quirós},
  title = {P2PaLA: Page to PAGE Layout Analysis tookit},
  year = {2017},
  publisher = {GitHub},
  note = {GitHub repository},
  howpublished = {\url{https://github.com/lquirosd/P2PaLA}},
}

Check this paper for more details Arxiv.

Requirements

  • Linux (OSX may work, but untested.).
  • Python (2.7, 3.6 under conda virtual environment is recomended)
  • Numpy
  • PyTorch (1.0). PyTorch 0.3.1 compatible on this branch
  • OpenCv (3.4.5.20).
  • NVIDIA GPU + CUDA CuDNN (CPU mode and CUDA without CuDNN works, but is not recomended for training).
  • tensorboard-pytorch (v0.9) [Optional]. pip install tensorboardX > A diferent conda env is recomended to keep tensorflow separated from PyTorch

Install

python setup.py install

To install python dependencies alone, use requirements file conda env create --file conda_requirements.yml

Usage

  1. Input data must follow the folder structure data_tag/page, where images must be into the data_tag folder and xml files into page. For example:
mkdir -p data/{train,val,test,prod}/page;
tree data;
data
├── prod
│   ├── page
│   │   ├── prod_0.xml
│   │   └── prod_1.xml
│   ├── prod_0.jpg
│   └── prod_1.jpg
├── test
│   ├── page
│   │   ├── test_0.xml
│   │   └── test_1.xml
│   ├── test_0.jpg
│   └── test_1.jpg
├── train
│   ├── page
│   │   ├── train_0.xml
│   │   └── train_1.xml
│   ├── train_0.jpg
│   └── train_1.jpg
└── val
    ├── page
    │   ├── val_0.xml
    │   └── val_1.xml
    ├── val_0.jpg
    └── val_1.jpg
  1. Run the tool.
python P2PaLA.py --config config.txt --tr_data ./data/train --te_data ./data/test --log_comment "_foo"

❗ Pre-trained models available here

  1. Use TensorBoard to visualize train status:
tensorboard --logdir ./work/runs
  1. xml-PAGE files must be at "./work/results/test/"

We recommend Transkribus or nw-page-editor to visualize and edit PAGE-xml files.

  1. For detail about arguments and config file, see docs or python P2PaLA.py -h.
  2. For more detailed example see egs:
    • Bozen dataset see
    • cBAD complex competition dataset see
    • OHG dataset see

License

GNU General Public License v3.0 See LICENSE to see the full text.

Acknowledgments

Code is inspired by pix2pix and pytorch-CycleGAN-and-pix2pix