Semantic Table Detection with LayoutLMv3

TL;DR

This study applies LayoutLMv3 to semantic table detection in financial documents but finds no significant improvement, highlighting challenges in leveraging semantic info for table detection.

Contribution

The paper evaluates LayoutLMv3 for semantic table detection and discusses potential reasons for its limited effectiveness in this task.

Findings

No improvement in table detection accuracy with semantic info

Possible issues with model weights or hyperparameter tuning

Semantic information may not enhance table detection performance

Abstract

This paper presents an application of the LayoutLMv3 model for semantic table detection on financial documents from the IIIT-AR-13K dataset. The motivation behind this paper's experiment was that LayoutLMv3's official paper had no results for table detection using semantic information. We concluded that our approach did not improve the model's table detection capabilities, for which we can give several possible reasons. Either the model's weights were unsuitable for our purpose, or we needed to invest more time in optimising the model's hyperparameters. It is also possible that semantic information does not improve a model's table detection accuracy.