SciPostLayout: A Dataset for Layout Analysis and Layout Generation of Scientific Posters
Shohei Tanaka, Hao Wang, Yoshitaka Ushiku
TL;DR
SciPostLayout presents the first large, publicly licensed dataset of scientific posters (7,855 posters) with manual nine-category layout annotations and 100 associated papers, enabling rigorous evaluation of layout analysis and poster-generation methods. The authors benchmark mainstream layout-analysis models (e.g., LayoutLMv3, DiT) and layout-generation approaches (e.g., LayoutDM, LayoutPrompter) and show posters pose greater challenges than scientific papers due to typography and panel variability. They also explore GPT-4-based paper-to-poster generation in Gen-T and Gen-P settings, finding that while LLMs can produce aligned layouts with reduced overlap, achieving layouts closely matching real posters remains difficult. The dataset, public licenses, and baseline results establish a foundation for developing and evaluating automatic poster generation systems with practical, CC-BY-licensed data, guiding future improvements in layout understanding and generation.
Abstract
Scientific posters are used to present the contributions of scientific papers effectively in a graphical format. However, creating a well-designed poster that efficiently summarizes the core of a paper is both labor-intensive and time-consuming. A system that can automatically generate well-designed posters from scientific papers would reduce the workload of authors and help readers understand the outline of the paper visually. Despite the demand for poster generation systems, only a limited research has been conduced due to the lack of publicly available datasets. Thus, in this study, we built the SciPostLayout dataset, which consists of 7,855 scientific posters and manual layout annotations for layout analysis and generation. SciPostLayout also contains 100 scientific papers paired with the posters. All of the posters and papers in our dataset are under the CC-BY license and are publicly available. As benchmark tests for the collected dataset, we conducted experiments for layout analysis and generation utilizing existing computer vision models and found that both layout analysis and generation of posters using SciPostLayout are more challenging than with scientific papers. We also conducted experiments on generating layouts from scientific papers to demonstrate the potential of utilizing LLM as a scientific poster generation system. The dataset is publicly available at https://huggingface.co/datasets/omron-sinicx/scipostlayout_v2. The code is also publicly available at https://github.com/omron-sinicx/scipostlayout.