3D face alignment / reconstruction
Official PyTorch implementation of 3DDFA_V2 for fast, accurate and stable 3D dense face alignment from a single image, producing 3DMM parameters and dense mesh.
Solves the same problem with a different architecture or at a different layer. Expect to rewrite the integration.
Choose if you need a full face analysis ecosystem with detection and recognition, but it's a heavier framework and not a drop-in for dense 3D alignment.
State-of-the-art 2D and 3D Face Analysis Project
Choose if you want a single API for many face tasks including face mesh, but it's a high-level wrapper, not a standalone 3D alignment model.
UniFace: A Unified Face Analysis Library for Python | Detection, alignment, landmarks, face-mesh, recognition, parsing, gaze, attributes and anti-spoofing under one API.
Choose if you want a 3DMM+landmark fusion approach, but it's a research codebase not a production library.
3DV 2021: Synergy between 3DMM and 3D Landmarks for Accurate 3D Facial Geometry
Choose if you need a model trained on a diverse 3D head dataset, but it's a dataset/model, not a drop-in library.
Official repo for DAD-3DHeads: A Large-scale Dense, Accurate and Diverse Dataset for 3D Head Alignment from a Single Image (CVPR 2022).
Choose if you need FLAME-based head tracking and reconstruction, but it's offline and not real-time.
FLAME head tracker for single image reconstruction and monocular video tracking. [Note: This tracker operates offline and is not intended for real-time applications.]
Choose if you need multi-head 3D alignment from a single image, but it's specialized for head detection/pose rather than dense face mesh.
Official repo for VGGHeads: 3D Multi Head Alignment with a Large-Scale Synthetic Dataset..
Choose if you need a C++ 3DMM fitting library with permissive license, but it's not a turnkey Python model.
A lightweight 3D Morphable Face Model library in modern C++
Choose if you want a recent single-image 3D face reconstruction method with screen-space priors, but it's a research codebase.
[Official Code] Pixel3DMM: Versatile Screen-Space Priors for Single-Image 3D Face Reconstruction
Choose if you need metrically accurate 3D face reconstruction, but it's a different approach and not real-time.
MICA - Towards Metrical Reconstruction of Human Faces [ECCV2022]
An earlier or more limited way to do the job. Still the right call when you need something small, proven, or CPU-only.
Choose if you need 6DoF pose and face alignment in one model, but it's from 2021 and not dense 3D mesh.
The official PyTorch implementation of img2pose: Face Alignment and Detection via 6DoF, Face Pose Estimation - CVPR 2021
Choose if you need a proven 2018 method for dense 3D face alignment, but it's unmaintained and not real-time.
Joint 3D Face Reconstruction and Dense Alignment with Position Map Regression Network (ECCV 2018)
Choose if you prefer optimization-based 3DMM fitting, but it's older and not real-time.
A 3DMM fitting framework using Pytorch.
Choose if you need a simple older deep learning approach for 3D face reconstruction, but it's from 2018 and unmaintained.
Reconstructing real-time 3D faces from 2D images using deep learning.
Choose if you want a C++ deep learning 3DMM regression method, but it's from 2017 and unmaintained.
Regressing Robust and Discriminative 3D Morphable Models with a very Deep Neural Network
Choose if you need a simple dlib+3DMM fitting pipeline, but it's from 2016 and unmaintained.
Facial Landmark Detection and head pose compute use dlib, Real time Face Reconstruction use 3D Morphable Face Model fitting
Choose if you need a Matlab 3DMM edge-fitting implementation, but it's from 2015 and not Python.
Matlab implementation of algorithm for fitting 3D Morphable Model to images using edge features
These projects were analysed and named 3DDFA_V2 among their alternatives. The relationship is not symmetric — how 3DDFA_V2 rates them is a separate judgement, made when 3DDFA_V2 is analysed in its own right.
calls 3DDFA_V2 “dense 3D face alignment with full-pose support; pick when you need 3D mesh rather than sparse landmarks.”
:fire: 2D and 3D Face alignment library build using pytorch
calls 3DDFA_V2 “A specialized 3D dense face alignment implementation, narrower than InsightFace.”
State-of-the-art 2D and 3D Face Analysis Project