The authors worked with 51 professional UI designers to organize systematic aesthetic principles and construct a shared data foundation of 1,395 executable web UIs.
Multimodal models show a gap between aesthetic judgment and design action for UIs
A new benchmark reveals that even strong models achieve only moderate aesthetic quality, and correct judgments do not consistently lead to successful repairs.
Chinese Tech
Zhijie Deng · Ling Li · Junhao Ji · Siwei Lyu · Zhipeng Xu · Zulong Chen · +4 more
The Hong Kong University of Science and Technology (Guangzhou) · Alibaba Group · The Hong Kong University of Science and Technology
Research Digest··2 min read
The authors introduce AUV-Bench, a benchmark of 1,395 executable web UIs with four tasks: aesthetic scoring, diagnosis, repair, and text-to-UI generation.
Why this paper
From Alibaba Group and 2 others
In one line
Multimodal models show moderate aesthetic scoring but limited fine-grained diagnosis and a gap between judging and fixing UI problems.
What we could check
- ·No code link found
- ·No weights link found
- ·No dataset link found
- ·No compute details found
- ·No stated limitations found
- ✓Reports numbers on named benchmarks
Observed from the paper text and links we have. Absence here means we did not find it, not that it does not exist.
§