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.

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 - RoBERTa-base2023-05-02

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.

Ranking Table

Description Paper Source Code
DateMethodF1APPrecisionRecall
2023-05-25GraphDoc+Classify+Merge65.93%44.45%76.43%57.97%
2023-05-02baseline - RoBERTa-base with synthetic pre-training63.12%54.94%60.89%65.52%
2023-05-02baseline - RoBERTa-base59.76%50.36%56.85%62.99%
2023-05-02baseline - LayoutLMv3 with unsupervised and synthetic pre-training58.58%48.89%57.97%59.19%
2023-05-02baseline - LayoutLMv3 with unsupervised pre-training54.18%37.51%54.89%53.48%
2023-05-24YOLOv8X+Grid40.19%19.61%51.41%32.99%
2023-05-25SRCB Submission on Line Item Recognition35.13%13.02%36.44%33.91%

Ranking Graphic

Ranking Graphic