Table detection in business document images by message passing networks

作者：

Highlights：

• A table detection approach with heterogeneous formats for business documents working on anonymized data.

• A new graph neural network architecture that poses the table detection problems in terms of node and edge classification.

• A final consensus layer based on the belief propagation algorithm to marginalize the edge probability.

• Extensive experimentation on three document datasets.

摘要

•A table detection approach with heterogeneous formats for business documents working on anonymized data.•A new graph neural network architecture that poses the table detection problems in terms of node and edge classification.•A final consensus layer based on the belief propagation algorithm to marginalize the edge probability.•Extensive experimentation on three document datasets.

论文关键词：Business document processing,Anonymized document processing,Table detection,Graph neural networks,Node and edge classification

论文评审过程：Received 31 December 2020, Revised 21 December 2021, Accepted 8 March 2022, Available online 11 March 2022, Version of Record 15 March 2022.

论文官网地址：https://doi.org/10.1016/j.patcog.2022.108641