Researchers have developed PosterVisor, a new framework designed to improve the generation of scientific posters from multimodal papers. Unlike previous methods that used transient prompts and isolated stage validation, PosterVisor employs persistent control through Semantic-Geometric Contracts (SGCs). These contracts bind paper content and visual assets to specific requirements, with only fully instantiated records becoming executable assertions. The system uses Recursive Contract Enforcement (RCE) to dynamically check stages and re-verify affected checkpoints during repairs, preventing regressions. Evaluations on the Paper2Poster benchmark show PosterVisor-PPT significantly improves poster-grounded QA accuracy and is preferred by human judges over existing methods. AI
IMPACT This framework could improve the efficiency and quality of scientific communication by automating poster creation with better accuracy and user preference.
RANK_REASON The cluster contains a research paper detailing a new method and framework for scientific poster generation. [lever_c_demoted from research: ic=1 ai=1.0]
- Paper2Poster
- PosterGen
- PosterVisor
- PosterVisor-PPT
- Recursive Contract Enforcement
- Semantic-Geometric Contract
- vision-language model
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
