method: GraphDoc+Classify+Merge2023-05-25
Authors: Yan Wang, Jiefeng Ma, Zhenrong Zhang, Pengfei Hu, Jianshu Zhang, Jun Du
Affiliation: University of Science and Technology of China (USTC), iFLYTEK AI Research
Description: We pre-trained several GraphDoc models on provided unlabelled documents under different configurations. We then fine-tuned the models on the training set for 500-1000 epochs. After classifying OCR boxes into various categories, we proposed a Merger module to handle the aggregation process.
We also used some pre/post-processing according to the text content and distances between OCR boxes. Finally, we adopted model ensembling to further enhance the system performance.
method: baseline - RoBERTa-base with synthetic pre-training2023-05-02
Authors: Organizers
Affiliation: Rossum.ai, Czech Technical University in Prague, University of La Rochelle
Description: Baseline method. Uses multi-label NER formulation with RoBERTa base as the backbone. It is pre-trained on the synthetic part of the DocILE dataset.
method: baseline - LayoutLMv3 with unsupervised and synthetic pre-training2023-05-02
Authors: Organizers
Affiliation: Rossum.ai, Czech Technical University in Prague, University of La Rochelle
Description: Baseline method. Uses multi-label NER formulation with LayoutLMv3 as the backbone. It is pre-trained on the unlabelled and synthetic parts of the DocILE dataset.
Date | Method | F1 | AP | Precision | Recall | |||
---|---|---|---|---|---|---|---|---|
2023-05-25 | GraphDoc+Classify+Merge | 75.93% | 57.89% | 80.82% | 71.60% | |||
2023-05-02 | baseline - RoBERTa-base with synthetic pre-training | 69.82% | 58.28% | 70.98% | 68.71% | |||
2023-05-02 | baseline - LayoutLMv3 with unsupervised and synthetic pre-training | 69.06% | 58.23% | 70.95% | 67.27% | |||
2023-05-02 | baseline - LayoutLMv3 with unsupervised pre-training | 66.12% | 53.12% | 68.22% | 64.14% |