SciPostLayout: A Dataset for Layout Analysis and Layout Generation of Scientific Posters

Hao Wang, Shohei Tanaka, Yoshitaka Ushiku; Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops, 2024, pp. 8136-8141

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 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 generating posters from papers has a limited research population due to the lack of publicly available datasets. Thus in this paper we built the SciPostLayout dataset which consists of 7855 scientific posters and manual layout annotations for layout analysis and generation. All of the posters in our dataset are under the CC-BY license and publicly available. As benchmark tests for the collected dataset we conducted experiments for layout analysis and generation using existing computer vision models. Experimental results show that layout analysis and generation of posters using SciPostLayout are more challenging than with scientific papers. The dataset is publicly available at https://huggingface.co/datasets/omron-sinicx/scipostlayout_v1.

Related Material


[pdf] [arXiv]
[bibtex]
@InProceedings{Wang_2024_CVPR, author = {Wang, Hao and Tanaka, Shohei and Ushiku, Yoshitaka}, title = {SciPostLayout: A Dataset for Layout Analysis and Layout Generation of Scientific Posters}, booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops}, month = {June}, year = {2024}, pages = {8136-8141} }