Synthetic Data Augmentation for Table Detection: Re-evaluating TableNet's Performance with Automatically Generated Document Images

2025-06-17

AI Summary

To address the high annotation cost and poor generalization in document table detection, this paper proposes the first end-to-end controllable LaTeX-driven synthetic framework, enabling realistic two-column document image generation with adjustable layout, styling, and resolution, alongside precise pixel-level table masks. The framework integrates geometric layout randomization, high-fidelity rendering, and the TableNet segmentation model to support systematic model training and evaluation without real annotated data. On synthetic test sets, it achieves XOR errors of 4.04% (256×256) and 4.33% (1024×1024); on the real-world Marmot benchmark, it attains 9.18%, significantly outperforming prior methods. This work pioneers the deep integration of controllable synthesis with pixel-level evaluation, establishing a high-quality synthetic data paradigm and a reproducible evaluation benchmark for document understanding.

Technology Category

• Natural Language Processing: Code Generation / Program Synthesis from Natural Language
• Computer Vision: Computational Photography, Image & Video Synthesis
• Machine Learning: Large Multimodal Models (LMMs)

Application Category

• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

Document pages captured by smartphones or scanners often contain tables, yet manual extraction is slow and error-prone. We introduce an automated LaTeX-based pipeline that synthesizes realistic two-column pages with visually diverse table layouts and aligned ground-truth masks. The generated corpus augments the real-world Marmot benchmark and enables a systematic resolution study of TableNet. Training TableNet on our synthetic data achieves a pixel-wise XOR error of 4.04% on our synthetic test set with a 256x256 input resolution, and 4.33% with 1024x1024. The best performance on the Marmot benchmark is 9.18% (at 256x256), while cutting manual annotation effort through automation.

Problem

• Automating table detection in document images
• Reducing manual annotation effort for table extraction
• Evaluating TableNet's performance with synthetic data

Innovation

• Automated LaTeX-based synthetic data pipeline
• Generates diverse table layouts with ground-truth
• Reduces manual annotation with synthetic augmentation