GenPlot: Increasing the Scale and Diversity of Chart Derendering Data

TL;DR

GenPlot is a synthetic data generator that significantly expands the scale and diversity of chart derendering datasets, enhancing the training of visual language models for understanding various plot types.

Contribution

It introduces a scalable plot generator that creates billions of synthetic charts, addressing dataset limitations in chart derendering tasks.

Findings

Generated billions of diverse synthetic plots

Improved performance on chart derendering benchmarks

Enhanced model robustness across plot types

Abstract

Vertical bars, horizontal bars, dot, scatter, and line plots provide a diverse set of visualizations to represent data. To understand these plots, one must be able to recognize textual components, locate data points in a plot, and process diverse visual contexts to extract information. In recent works such as Pix2Struct, Matcha, and Deplot, OCR-free chart-to-text translation has achieved state-of-the-art results on visual language tasks. These results outline the importance of chart-derendering as a pre-training objective, yet existing datasets provide a fixed set of training examples. In this paper, we propose GenPlot; a plot generator that can generate billions of additional plots for chart-derendering using synthetic data.