analiza aiqbench
This research introduces MAVEN, a multi-agent framework that significantly enhances cultural fidelity in text-to-video generation by decomposing prompts into specific dimensions like person, action, and location. By evaluating both mono-cultural and cross-cultural scenarios across three distinct cultures, the study demonstrates how parallel agent specialization improves visual relevance without compromising quality or temporal consistency. The release of a new benchmark comprising hundreds of culturally grounded videos provides essential data for advancing equitable AI video synthesis.
When Cultures Move: Measuring and Improving Multicultural Text-to-Video Generation
arXiv:2605.16716v5 Announce Type: replace-cross Abstract: Text-to-video (T2V) generation has rapidly progressed in visual fidelity, yet its ability to faithfully represent multiple cultures within a single prompt remains underexplored. We introduce MAVEN, a multi-agent prompt refinement framework designed to improve cultural fidelity in both mono-cultural and cross-cultural T2V generation. MAVEN decomposes prompts into person, action, and location dimensions, handled by specialized agents operating in parallel or sequentially. To support systematic evaluation, we contribute a new benchmark of 243 culturally grounded prompts and 972 corresponding videos, spanning three cultures (Chinese, American, Romanian), three action categories, and both mono-cultural and cross-cultural scenarios. Evaluations combining CLIP-based metrics, VLM-as-judge assessments, and videoquality measures show that multi-agent refinement, particularly parallel specialization, significantly improves cultural relevance while preserving visual quality and temporal consistency. The dataset and code are available at https://github.com/AIM-SCU/MAVEN
Abstract:Text-to-video (T2V) generation has rapidly progressed in visual fidelity, yet its ability to faithfully represent multiple cultures within a single prompt remains underexplored. We introduce MAVEN, a multi-agent prompt refinement framework designed to improve cultural fidelity in both mono-cultural and cross-cultural T2V generation. MAVEN decomposes prompts into person, action, and location dimensions, handled by specialized agents operating in parallel or sequentially. To support systematic evaluation, we contribute a new benchmark of 243 culturally grounded prompts and 972 corresponding videos, spanning three cultures (Chinese, American, Romanian), three action categories, and both mono-cultural and cross-cultural scenarios. Evaluations combining CLIP-based metrics, VLM-as-judge assessments, and videoquality measures show that multi-agent refinement, particularly parallel specialization, significantly improves cultural relevance while preserving visual quality and temporal consistency. The dataset and code are available at this https URL
[14] pages, [6] figures, [11] tables, appendix included. Preprint
Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
arXiv:2605.16716 [cs.CV]
(or arXiv:2605.16716v5 [cs.CV] for this version)
Text integral sub licența deschisă a publicației · 157 cuvinte
Subiecte conexe
Dyna-2: General Purpose Physical Agents with Hour-Long Autonomy — Jason Ma, Dyna Robotics
Latent Space · 13 oct. 2026
Microsoft’s AI chief says superintelligence is near, but won’t take your job
The Verge · 8 iun. 2026
MAVEN A Multi-Agent Framework for Multicultural Text-to-Video Generation
arXiv · 6 iun. 2026
Humanity’s Last Invention — Richard Socher of Recursive
Latent Space · 14 sept. 2026