Optimized Table Tokenization for Table Structure Recognition

TL;DR

This paper introduces OTSL, an optimized table-structure language that reduces token count and sequence length, leading to faster, more accurate table structure recognition from images with minimal post-processing.

Contribution

The paper proposes OTSL, a new minimal vocabulary language for table-structure recognition that improves accuracy and efficiency over traditional HTML-based representations.

Findings

Token count reduced from 28+ to 5

Sequence length halved on average

Inference time is halved

Abstract

Extracting tables from documents is a crucial task in any document conversion pipeline. Recently, transformer-based models have demonstrated that table-structure can be recognized with impressive accuracy using Image-to-Markup-Sequence (Im2Seq) approaches. Taking only the image of a table, such models predict a sequence of tokens (e.g. in HTML, LaTeX) which represent the structure of the table. Since the token representation of the table structure has a significant impact on the accuracy and run-time performance of any Im2Seq model, we investigate in this paper how table-structure representation can be optimised. We propose a new, optimised table-structure language (OTSL) with a minimized vocabulary and specific rules. The benefits of OTSL are that it reduces the number of tokens to 5 (HTML needs 28+) and shortens the sequence length to half of HTML on average. Consequently, model accuracy improves significantly, inference time is halved compared to HTML-based models, and the predicted table structures are always syntactically correct. This in turn eliminates most post-processing needs.