VGBench: Evaluating Large Language Models on Vector Graphics Understanding and Generation

TL;DR

VGBench is a comprehensive benchmark designed to evaluate large language models' abilities in understanding and generating vector graphics, covering various formats, question types, and prompting techniques, with insights into their strengths and limitations.

Contribution

This work introduces VGBench, the first extensive benchmark for LLMs on vector graphics, encompassing diverse formats, tasks, and evaluation methods, filling a gap in current research.

Findings

LLMs perform well on understanding and generation tasks.

Performance drops on low-level formats like SVG.

Benchmark and data will be open-sourced for community use.

Abstract

In the realm of vision models, the primary mode of representation is using pixels to rasterize the visual world. Yet this is not always the best or unique way to represent visual content, especially for designers and artists who depict the world using geometry primitives such as polygons. Vector graphics (VG), on the other hand, offer a textual representation of visual content, which can be more concise and powerful for content like cartoons, sketches and scientific figures. Recent studies have shown promising results on processing vector graphics with capable Large Language Models (LLMs). However, such works focus solely on qualitative results, understanding, or a specific type of vector graphics. We propose VGBench, a comprehensive benchmark for LLMs on handling vector graphics through diverse aspects, including (a) both visual understanding and generation, (b) evaluation of various vector graphics formats, (c) diverse question types, (d) wide range of prompting techniques, (e) under multiple LLMs and (f) comparison with VLMs on rasterized representations. Evaluating on our collected 4279 understanding and 5845 generation samples, we find that LLMs show strong capability on both aspects while exhibiting less desirable performance on low-level formats (SVG). Both data and evaluation pipeline will be open-sourced at https://vgbench.github.io.