Self-Supervised Pre-Training for Table Structure Recognition Transformer

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Peng, ShengYun, Lee, Seongmin, Wang, Xiaojing, Balasubramaniyan, Rajarajeswari, Chau, Duen Horng
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913243789787136
author Peng, ShengYun
Lee, Seongmin
Wang, Xiaojing
Balasubramaniyan, Rajarajeswari
Chau, Duen Horng
author_facet Peng, ShengYun
Lee, Seongmin
Wang, Xiaojing
Balasubramaniyan, Rajarajeswari
Chau, Duen Horng
contents Table structure recognition (TSR) aims to convert tabular images into a machine-readable format. Although hybrid convolutional neural network (CNN)-transformer architecture is widely used in existing approaches, linear projection transformer has outperformed the hybrid architecture in numerous vision tasks due to its simplicity and efficiency. However, existing research has demonstrated that a direct replacement of CNN backbone with linear projection leads to a marked performance drop. In this work, we resolve the issue by proposing a self-supervised pre-training (SSP) method for TSR transformers. We discover that the performance gap between the linear projection transformer and the hybrid CNN-transformer can be mitigated by SSP of the visual encoder in the TSR model. We conducted reproducible ablation studies and open-sourced our code at https://github.com/poloclub/unitable to enhance transparency, inspire innovations, and facilitate fair comparisons in our domain as tables are a promising modality for representation learning.
format Preprint
id arxiv_https___arxiv_org_abs_2402_15578
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Self-Supervised Pre-Training for Table Structure Recognition Transformer
Peng, ShengYun
Lee, Seongmin
Wang, Xiaojing
Balasubramaniyan, Rajarajeswari
Chau, Duen Horng
Computer Vision and Pattern Recognition
Table structure recognition (TSR) aims to convert tabular images into a machine-readable format. Although hybrid convolutional neural network (CNN)-transformer architecture is widely used in existing approaches, linear projection transformer has outperformed the hybrid architecture in numerous vision tasks due to its simplicity and efficiency. However, existing research has demonstrated that a direct replacement of CNN backbone with linear projection leads to a marked performance drop. In this work, we resolve the issue by proposing a self-supervised pre-training (SSP) method for TSR transformers. We discover that the performance gap between the linear projection transformer and the hybrid CNN-transformer can be mitigated by SSP of the visual encoder in the TSR model. We conducted reproducible ablation studies and open-sourced our code at https://github.com/poloclub/unitable to enhance transparency, inspire innovations, and facilitate fair comparisons in our domain as tables are a promising modality for representation learning.
title Self-Supervised Pre-Training for Table Structure Recognition Transformer
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2402.15578