2026 · Conference accepted

From Visual Widgets to UI Code: Efficient Tool-Grounded Generation

WidgetGen reconstructs widget screenshots as React JSX. OCR text boxes and color palettes guide whole-widget layout reasoning and optional chart generation before one terminal render; generated render/code pairs also support model fine-tuning.

Scope and limitations

The main comparison uses six models and 1,000 test widgets from one dataset. Its nine image-based metrics do not test interaction, accessibility or maintainability. The supervision experiment covers six Qwen-family checkpoints; broader UI and model-family generalization remains open.

Why it fits this collection

Studies evidence-assisted recovery of executable interface code from compact visual references.

Publication and review details

Venue
MUCG @ ECCV 2026
Status
Conference accepted
Publication / posting date
2026 (exact date not verified)
Date source
The official 2026 MUCG workshop program confirms inclusion; a proceedings publication date was not verified.
First arXiv submission
August 12, 2026
arXiv
2608.12611
Reviewed arXiv version
v1
DOI
10.48550/arXiv.2608.12611

The official MUCG 2026 program lists this paper; proceedings publication was not verified, and OpenReview required browser verification. The arXiv paper is CC BY 4.0. No dedicated implementation or weight link was found in the reviewed paper/program; Widget2Code is the separate benchmark source. Nothing was executed.

Sources reviewed

Source review only; experiments were not rerun for this collection. Cite the original authors using the linked paper or BibTeX.

Primary sources

  1. https://arxiv.org/abs/2608.12611v1
  2. https://arxiv.org/html/2608.12611v1
  3. https://mllm-mucg.github.io/ECCV2026/