Towards Hierarchical Structure Understanding of Newspaper Images

Source

• Source name: arXiv
• Source kind: paper
• Authors / organization: William Mocaër, Solène Tarride, Thomas Constum, Merveilles Agbeti-Messan, Tom Simon, Clément Chatelain, Stéphane Nicolas, Pierrick Tranouez, Sébastien Cretin, Thierry Paquet
• Venue: arXiv: cs.CV
• Published: 2026-07-16
• Collected: 2026-07-18
• URL: http://arxiv.org/abs/2607.15082v1
• PDF: https://arxiv.org/pdf/2607.15082v1
• DOI: UNKNOWN

Plain-Language Brief

The paper presents two approaches for understanding the complex and nested hierarchical structures in newspaper images. The first approach is a modular pipeline that combines existing open-source models to detect layouts, predict reading order, and segment articles. The second approach introduces Tiramisu, a novel end-to-end transformer-based architecture that models document hierarchy through an iterative tiered process for various tasks such as section and article separation, block localization, semantic categorization, and reading order prediction. A new dataset called Finlam La Liberté is also released for evaluating hierarchical information retrieval in historical newspapers.

What the Document Says

• The paper addresses the challenge of understanding complex, nested hierarchical structures in newspaper images due to their dense, heterogeneous layouts.
• A modular bottom-up pipeline is presented that combines state-of-the-art open-source models for layout detection, reading order prediction, and article segmentation.
• Tiramisu, a novel end-to-end transformer-based architecture, is introduced to model document hierarchy through an iterative tiered process.
• Tiramisu performs various tasks including section and article separation, block localization, semantic categorization, and reading order prediction using attention mechanisms.
• Finlam La Liberté, a new dataset for evaluating hierarchical information retrieval in historical newspapers, is released.
• Experimental results demonstrate the effectiveness of both approaches in reconstructing complex newspaper hierarchies.

Key Concepts Mentioned

• newspaper images
• complex hierarchical structures
• bottom-up pipeline
• Tiramisu
• transformer-based architecture
• Finlam La Liberté
• experimental results

Claims Made by the Document

• Understanding newspaper images remains a challenging task.
• The proposed approaches are effective in reconstructing complex newspaper hierarchies.

Evidence Used

• The paper describes the complexity of newspaper images and presents two approaches for addressing this challenge.
• Experimental results demonstrate the effectiveness of both approaches.

Methods / Approach

• A modular bottom-up pipeline combining state-of-the-art open-source models
• Tiramisu, a novel end-to-end transformer-based architecture

Limitations Stated by the Authors

UNKNOWN

Limitations Noted by the Agent

Agent interpretation — not from the source.

UNKNOWN

Open Questions

UNKNOWN

Notes for Future Q&A

May include interpretation. Labeled where it does.

UNKNOWN

Raw Abstract / Excerpt

Understanding newspaper images remains a challenging task due to their complex, nested hierarchical structures and dense, heterogeneous layouts. In this paper, we explore two complementary approaches for newspaper structure understanding. First, we present a modular bottom-up pipeline that combines state-of-the-art open-source models: YOLO for layout detection, LayoutReader for reading order prediction, and a custom algorithm for article segmentation. This approach leverages existing robust components while maintaining flexibility and interpretability. Second, we introduce Tiramisu (Tiered Transformers for Hierarchical Structure Understanding), a novel end-to-end transformer-based architecture that explicitly models document hierarchy through an iterative tiered process. Tiramisu performs section and article separation, block localization, semantic categorization, and reading order prediction using highly parallelized attention mechanisms. Finally, we release Finlam La Liberté, a new dataset designed specifically for evaluating hierarchical information retrieval in historical newspapers. Experimental results demonstrate the effectiveness of both approaches in reconstructing complex newspaper hierarchies, with comparative analysis highlighting their respective strengths for scalable document digitization. The Tiramisu training code, including the synthetic newspaper generator, is available at https://git.litislab.fr/tiramisu/tiramisu-newspaper-articles-extractor.