SVG-Score: Human-Aligned Evaluation of Text-to-SVG Generation

arXiv:2609.03806 · cs.AI, cs.CV · Submitted 2026-09-03 · Read on arXiv

cs.AI, cs.CV

Submitted: 2026-09-03

Updated: 2026-09-03

Project page: https://potpov.github.io/svg-score-webpage

License: http://creativecommons.org/licenses/by/4.0/

The gist: Scalable Vector Graphics (SVG) generation is attracting increasing attention as generative models improve in expressiveness and controllability.

Terminology

Abstract

Scalable Vector Graphics (SVG) generation is attracting increasing attention as generative models improve in expressiveness and controllability. Progress, however, is held back by the lack of domain-specific evaluation protocols: current practice relies on metrics designed for natural images, most notably CLIPScore, which was never trained on vector graphics and aligns only partially with human judgment. We introduce, a human-aligned evaluation framework for text-to-SVG generation. Through controlled caption and image perturbations, we first show that CLIP-based scores barely react to the errors SVG generators actually make, such as wrong colors, counts, and spatial relations, and that off-the-shelf Vision-Language Model (VLM) judges, while more sensitive, respond unevenly across error types and SVG styles. We then introduce a human-annotated dataset for Semantic Alignment, measuring how faithfully a generated SVG reflects its caption. Building on it, we develop two complementary evaluators: CLIP scorers adapted to vector graphics and then aligned to human preferences, for fast large-scale evaluation, and a VLM judge trained with supervised fine-tuning and reward-shaped reinforcement learning, for more expressive and interpretable assessment. Using both, we benchmark major open-source, commercial, and optimization-based SVG generators on an independent caption set.

Sources

Related papers