yfeng95/PRNet

Health v2 · maintenance only

Flags: no_releases

Adoption not part of the score

5013 stars · 932 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-29, confidence not recorded

PRNet is a Python/TensorFlow implementation of the ECCV 2018 Position Map Regression Network for joint 3D face reconstruction and dense alignment from a single image. It runs faster than real-time on GPU and supports applications like pose estimation, depth images, texture editing, and face swapping.

Use cases

• reconstruct a 3D face mesh from a single photo
• perform dense facial landmark alignment including visibility
• estimate 3D head pose from an image
• swap faces between photos
• generate depth maps of faces
• edit facial textures like eyes for selfie editing

When to choose

• you need end-to-end 3D face reconstruction without 3DMM fitting
• you need dense alignment with over 40K points rather than just 68 landmarks
• you need real-time performance on a GPU
• you want face swapping or texture editing capabilities

When to avoid

• you need a maintained library with recent updates or modern framework support
• you cannot use TensorFlow 1.x or Python 2.7 era dependencies
• you need multi-face reconstruction in a single image
• you only need lightweight 2D landmark detection

Facets

library · maturity maintenance

machine-learning computer-vision image-processing deep-learning computer-vision image-processing deep-learning artificial-intelligence python 3d-face-reconstruction face-alignment face-swapping position-map-regression eccv-2018 tensorflow linux gpu

Member repositories

For agents

markdown · JSON · MCP: product_card(name="yfeng95/PRNet")