VisJudge-Bench: Aesthetics and Quality Assessment of Visualizations
This work addresses the challenge of assessing visualization aesthetics and quality for researchers and practitioners in data visualization and AI, though it is incremental as it builds on existing MLLM capabilities for aesthetic assessment.
The paper tackles the problem of evaluating visualization quality, which lacks a systematic benchmark for multimodal large language models (MLLMs), by introducing VisJudge-Bench, a comprehensive dataset with 3,090 expert-annotated samples, and shows that advanced MLLMs like GPT-5 have significant gaps compared to humans (MAE 0.551, correlation 0.429), while their proposed VisJudge model reduces MAE to 0.442 (19.8% improvement) and increases correlation to 0.681 (58.7% improvement).
Visualization, a domain-specific yet widely used form of imagery, is an effective way to turn complex datasets into intuitive insights, and its value depends on whether data are faithfully represented, clearly communicated, and aesthetically designed. However, evaluating visualization quality is challenging: unlike natural images, it requires simultaneous judgment across data encoding accuracy, information expressiveness, and visual aesthetics. Although multimodal large language models (MLLMs) have shown promising performance in aesthetic assessment of natural images, no systematic benchmark exists for measuring their capabilities in evaluating visualizations. To address this, we propose VisJudge-Bench, the first comprehensive benchmark for evaluating MLLMs' performance in assessing visualization aesthetics and quality. It contains 3,090 expert-annotated samples from real-world scenarios, covering single visualizations, multiple visualizations, and dashboards across 32 chart types. Systematic testing on this benchmark reveals that even the most advanced MLLMs (such as GPT-5) still exhibit significant gaps compared to human experts in judgment, with a Mean Absolute Error (MAE) of 0.551 and a correlation with human ratings of only 0.429. To address this issue, we propose VisJudge, a model specifically designed for visualization aesthetics and quality assessment. Experimental results demonstrate that VisJudge significantly narrows the gap with human judgment, reducing the MAE to 0.442 (a 19.8% reduction) and increasing the consistency with human experts to 0.681 (a 58.7% improvement) compared to GPT-5. The benchmark is available at https://github.com/HKUSTDial/VisJudgeBench.