Line Graphics Digitization: A Step Towards Full Automation

TL;DR

This paper introduces the Line Graphics dataset for detailed visual understanding of mathematical graphics, supporting segmentation and detection tasks, aiming to advance full automation in graphical document digitization.

Contribution

It presents a new dataset with pixel-wise annotations for mathematical graphics, enabling research on automatic graphical element recognition and analysis.

Findings

7 state-of-the-art models benchmarked on the dataset

Dataset covers 520 images from 450 documents across disciplines

Supports semantic segmentation and object detection tasks

Abstract

The digitization of documents allows for wider accessibility and reproducibility. While automatic digitization of document layout and text content has been a long-standing focus of research, this problem in regard to graphical elements, such as statistical plots, has been under-explored. In this paper, we introduce the task of fine-grained visual understanding of mathematical graphics and present the Line Graphics (LG) dataset, which includes pixel-wise annotations of 5 coarse and 10 fine-grained categories. Our dataset covers 520 images of mathematical graphics collected from 450 documents from different disciplines. Our proposed dataset can support two different computer vision tasks, i.e., semantic segmentation and object detection. To benchmark our LG dataset, we explore 7 state-of-the-art models. To foster further research on the digitization of statistical graphs, we will make the dataset, code, and models publicly available to the community.