Teklia/doc-ufcn-generic-page

Doc-UFCN - Generic page detection

The generic page detection model predicts single pages from document images.

Model description

The model has been trained using the Doc-UFCN library on Horae and READ-BAD datasets. It has been trained on images with their largest dimension equal to 768 pixels, keeping the original aspect ratio.

Evaluation results

The model achieves the following results:

dataset	set	IoU	F1	AP@[.5]	AP@[.75]	AP@[.5,.95]
HOME	test	93.92	95.84	98.98	98.98	97.61
Horae	test	96.68	98.31	99.76	98.49	98.08
Horae	test-300	95.66	97.27	98.87	98.45	97.38

How to use?

Please refer to the Doc-UFCN library page to use this model.

Cite us!

@inproceedings{doc_ufcn2021,
    author = {Boillet, Mélodie and Kermorvant, Christopher and Paquet, Thierry},
    title = {{Multiple Document Datasets Pre-training Improves Text Line Detection With
              Deep Neural Networks}},
    booktitle = {2020 25th International Conference on Pattern Recognition (ICPR)},
    year = {2021},
    month = Jan,
    pages = {2134-2141},
    doi = {10.1109/ICPR48806.2021.9412447}
}