uobinxiao / sparsetabledet

Table detection for visually rich document images

Paper Link

https://www.sciencedirect.com/science/article/abs/pii/S0950705123008304
https://arxiv.org/abs/2305.19181

Dataset

You can download the table detection dataset with the following link:

https://huggingface.co/datasets/uobinxiao/open_tables_icttd_for_table_detection

Pretrained Model

A pre-trained model can be downloaded here, which is trained by the OpenTables&ICT-TD dataset. It is worth mentioning that the merged training sets and the merged testing sets of OpenTables and ICT-TD are used for training and evaluation. The evaluation scores are as following:

MAP	AP50	AP55	AP60	AP65	AP70	AP75	AP80	AP85	AP90	AP95
0.954	0.980	0.978	0.977	0.977	0.973	0.969	0.964	0.954	0.933	0.837

MAR	AR50	AR55	AR60	AR65	AR70	AR75	AR80	AR85	AR90	AR95
0.980	0.999	0.998	0.998	0.997	0.995	0.994	0.991	0.985	0.966	0.881

Requirements

This codebase is built on top of Detectron2 and Sparse-RCNN. Follow the instructions here to install Detectron2.

Configuration and Training

Set config.yaml based on the dataset.
Set configs/icttd_opentables.res50.300pro.yaml to modify the model parameters and the output log directory.
Use python train_net.py to train the model.

Evaluation

Use python predict.py to evaluate the model.

python predict.py --input_dir <image_dir> --gt_json_path <path of the ground truth json file> --config-file <path of the config yaml file> --weight_path <path of the weight file>

Citing

Please cite our work if you think it is helpful:

@article{xiao2023table,
  title={Table detection for visually rich document images},
  author={Xiao, Bin and Simsek, Murat and Kantarci, Burak and Alkheir, Ala Abu},
  journal={Knowledge-Based Systems},
  volume={282},
  pages={111080},
  year={2023},
  publisher={Elsevier}
}

@article{xiao2023revisiting,
  title={Revisiting table detection datasets for visually rich documents},
  author={Xiao, Bin and Simsek, Murat and Kantarci, Burak and Alkheir, Ala Abu},
  journal={arXiv preprint arXiv:2305.04833},
  year={2023}
}