This repository provides the official PyTorch implementation of a 2020 ECCV paper on 3D dense face alignment. It uses 3D morphable models (3DMM) to reconstruct 3D facial geometry and landmarks from 2D images. The project includes face detection (FaceBoxes), 3DMM parameter regression, pose estimation, and mesh rendering. It supports ONNX export for faster inference and achieves real-time performance with optimizations for webcam-based applications.