metadata
license: cc-by-nc-sa-4.0
📝 Introduction The Q-Real dataset contains 10,000 images generated by multiple text-to-image models, covering a wide range of themes. For each image in the dataset, we provide the coordinates of the main entities. Additionally, for these entities, we include evaluation questions related to realism and plausibility, along with detailed descriptive analyses.