WISE: A World Knowledge-Informed Semantic Evaluation for Text-to-Image Generation
WISE: 一种基于世界知识的文本到图像生成语义评估方法
机构 * University of Science and Technology of China(中国科学技术大学)
专题命中 图像生成评测 :image generation(title,abstract);text-to-image(title,abstract);分类 cs.CV
AI总结 针对现有文本到图像生成模型缺乏复杂语义理解和世界知识整合评估的问题,提出WISE基准,包含25个子领域的1000个精心设计的提示,并引入WiScore指标评估知识-图像对齐,实验表明当前模型在整合世界知识方面存在显著局限。
Comments Accepted to ICML 2026. We have also released an updated version of the benchmark, WISE_Verified. Please refer to https://github.com/PKU-YuanGroup/WISE for the latest version