FAPOS

Facial Appearance Prediction for Orthognathic Surgery with Diffusion Models

This repository provides the facial landmark index set used in the FAPOS framework, developed for orthognathic surgery-related facial analysis and prediction tasks.

📌 Description

The landmark indices shared here are derived from the 478-point facial annotation scheme defined by Google’s MediaPipe, and have been carefully refined in our study to focus on regions that are anatomically and clinically relevant to orthognathic surgery.

In particular:

• Redundant or unrelated regions (e.g., periocular areas, eyebrows) were excluded.
• Landmarks were grouped into five anatomical regions:
  • Upper Midline
  • Upper Left / Upper Right
  • Lower Midline
  • Lower Left / Lower Right

The index set includes:

• A total of 282 selected landmark indices (upper face 119 points, lower face 163 points)
• Format: plain text (.txt)

📁 File Contents

• fapos_landmark_indices.txt:
  Contains Python-style dictionary lists of landmark indices grouped by region.
• Landmark_figure.pdf:
  Figure showing the anatomical regions and selected landmarks.

📖 Citation

If you use this landmark set in your research or applications, please cite our paper:

[Full citation will go here when available]