Large Language Models for Page Stream Segmentation

TL;DR

This paper introduces TABME++, a new benchmark with OCR annotations, and evaluates large language models for page stream segmentation, showing decoder-based LLMs outperform smaller models and emphasizing OCR robustness.

Contribution

The paper presents TABME++, an enhanced benchmark for PSS with OCR data, and assesses LLMs, demonstrating their superiority over smaller models in document segmentation tasks.

Findings

Decoder-based LLMs outperform smaller multimodal encoders.

Robust OCR is crucial for effective page stream segmentation.

Insights into challenges and advancements in PSS research.

Abstract

Page Stream Segmentation (PSS) is an essential prerequisite for automated document processing at scale. However, research progress has been limited by the absence of realistic public benchmarks. This paper works towards addressing this gap by introducing TABME++, an enhanced benchmark featuring commercial Optical Character Recognition (OCR) annotations. We evaluate the performance of large language models (LLMs) on PSS, focusing on decoder-based models fine-tuned with parameter-efficient methods. Our results show that decoder-based LLMs outperform smaller multimodal encoders. Through a review of existing PSS research and datasets, we identify key challenges and advancements in the field. Our findings highlight the key importance of robust OCR, providing valuable insights for the development of more effective document processing systems.