Read the evidence.
Understand the method.
Primary research, our interpretation and the implementation status of each idea. Published results belong to the original authors’ models and datasets; citation does not validate our product.
TransFBP: What Whole-Face Attention Can Teach a Face Analyzer
Can an analyzer consider relationships across the face instead of judging every feature against a fixed ideal?
2 May 2026 · Read the research note →Peer-reviewed · International Journal of Multimedia Information RetrievalH-ARN: Why a Beauty Rating Needs More Than an Average
Can two photos receive the same average rating while people disagree very differently about them?
21 June 2026 · Read the research note →Conference paper · code and checkpoint workflow availableFPEM and LiveBeauty: A Candidate Baseline, Not a Ready-Made Verdict
What does an available specialist model add beyond asking a general multimodal model for a rating?
ICCV 2025 · Read the research note →arXiv preprint · not presented here as peer-reviewedDisFace3DNet: Explaining a Prediction Without Claiming a Cause
Can a model explain which appearance and photographic components entered its own prediction?
8 October 2026 · Read the research note →arXiv preprint · preregistered exploratory studyAI vs Human Attractiveness Ratings: Ranking Is Not Calibration
If an AI ranks photos similarly to people, does its 8 out of 10 mean what a human panel’s 8 means?
2 September 2026 · Read the research note →Peer-reviewed · Aesthetic Plastic SurgeryValidating Face Analysis: Independent People, Independent Ratings
What evidence would justify saying a face analyzer has been validated against human ratings?
26 May 2026 · Read the research note →From research to our tools
Specialist tools retain their published methods. Our homepage and general face analysis use the six-dimension evidence framework. Independently calibrated human-rating prediction remains a research objective.