yfeng95/DECA

Health v2 · maintenance only

Flags: no_releases no_license

Adoption not part of the score

2514 stars · 497 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

DECA is the official PyTorch implementation of a SIGGRAPH 2021 method that reconstructs a detailed 3D head model (pose, shape, facial detail, lighting) from a single input image using the FLAME face model. The reconstructed model can be animated with realistic wrinkle deformations across poses and expressions.

Use cases

• reconstruct a 3D face model from a single photo
• animate a face with expressions and poses from an image
• extract detailed facial geometry and wrinkles from photos
• build 3D avatars from unconstrained face images
• benchmark 3D face shape reconstruction accuracy
• export reconstructed head models as obj files
• estimate head pose and lighting from a face image

When to choose

• you need single-image 3D face reconstruction with fine detail
• you want animatable FLAME-based head models
• you need robust reconstruction under varied poses, lighting, and occlusions
• you are doing research on 3D face modeling or avatar creation

When to avoid

• you need a production-ready product rather than research code
• you cannot obtain the FLAME model license and trained weights
• you need full-body or multi-person reconstruction
• you require real-time performance on low-end hardware

Facets

library · maturity maintenance

machine-learning computer-vision image-processing deep-learning computer-vision graphics artificial-intelligence machine-learning python windows 3d-face-reconstruction flame-model face-animation pytorch single-image siggraph-2021 3d-head-modeling research-code gpu linux macos

Member repositories

For agents

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