LineCap: Line Charts for Data Visualization Captioning Models



Anita Mahinpei, Zona Kostic, Chris Tanner

 View presentation:2022-10-19T19:54:00ZGMT-0600Change your timezone on the schedule page
2022-10-19T19:54:00Z

Exemplar figure, described by caption below — Diagram showing a figure captioning system. The system is made of two boxes titled Line Count Prediction Model and Caption Generation Model. Arrows indicate that both boxes are taking a single-lined chart as input. The output of the Line Count Prediction Model is a number. This output is passed to the Caption Generation Model as input. The output of the Caption Generation Model is "Ylabel decreases at a decreasing rate".

Prerecorded Talk

The live footage of the talk, including the Q&A, can be viewed on the session page, Visualization Systems and Graph Visualization.

Keywords

figure captioning, line charts, deep learning dataset

Abstract

Data visualization captions help readers understand the purpose of a visualization and are crucial for individuals with visual impairments. The prevalence of poor figure captions and the successful application of deep learning approaches to image captioning motivate the use of similar techniques for automated figure captioning. However, research in this field has been stunted by the lack of suitable datasets. We introduce LineCap, a novel figure captioning dataset of 3,528 figures, and we provide insights from curating this dataset and using end-to-end deep learning models for automated figure captioning.