Annotated Dataset of Real and AI-Generated Faces with Dentofacial Markers

This dataset contains annotated images of both real human portraits and AI-generated faces labeled with a standardized set of dentofacial markers. The annotations are intended to support the detection of synthetic faces generated by diffusion-based text-to-image models. Each image has been evaluated for key dentofacial attributes such as tooth shape, gum anatomy, pupil symmetry, gaze direction, skin pore visibility, and others, based on the criteria defined in the accompanying annotation guideline. Annotations were performed manually by trained experts to ensure high accuracy and consistency. The dataset includes: Real face images sourced from the FFHQ dataset AI-generated faces from various diffusion models (e.g., SDXL, DALL·E 3, FLUX, MidJourney) Annotation metadata in JSON. This dataset serves as the empirical foundation for the study titled "Dentofacial Markers for the Detection of Faces Generated by Diffusion-Based Text-to-Image Artificial Intelligence Models" and is made publicly available to support further research in synthetic media detection, facial forensics, and AI accountability.