FaceDemo: a simple 3D face alignment and warping demo

Input: face image/video

Output: Textured 3D face model

Overview

This is the author's implementation of 3D face alignment and warping method described in:
Mirror Mirror: Crowdsourcing Better Portraits
Jun-Yan Zhu, Aseem Agarwala, Alexei A. Efros, Eli Shechtman, Jue Wang
In ACM Transactions on Graphics (Proceedings of SIGGRAPH Asia 2014)

Please cite our paper if you use our code for your research.

Given an image or a video sequence that contains a face, the program can align and warp the face to a 3D face model. The user can further rotate/translate the face, and adjust the camera projection (perspective projection or orthogonal projection). The program also provides different OpenGL rendering options. See "Usage" for details.

Contact: Jun-Yan Zhu (junyanz@berkeley.edu)

Requirements

• OpenCV:

sudo apt-get install libopencv-dev

• OpenGL and GLUT:

sudo apt-get install libglu1-mesa-dev freeglut3-dev mesa-common-dev
sudo apt-get install libxmu-dev libxi-dev

• CLM Tracker by Jason Saragih (included)

Getting Started (Linux)

• Download and unzip the code.

• Run the following command.

mkdir cbuild
cd cbuild
cmake -DCMAKE_BUILD_TYPE=Release  ../
make

Usage

1. Command line interface: ./FaceDemo mode filePath
   • Example (image): ./FaceDemo image ./data/bush.jpg
   • Example (video): ./FaceDemo video ./data/eli.avi
2. The program displays three windows:

• The "Detected Points" window shows the original input frame with the detected feature points.
• The "3D Face" window shows a 3D face model rendered by OpenGL.
• The "Projected Face" window shows a 2D projection result of the 3D face model. The program projects the 3D face back to the original image.

3. User Interaction: "3D Face" window takes the user's keyboard input.
   • shift + arrow keys: translate the face.
   • arrow keys: rotate the face.
   • +/-: zoom in/out
   • p: toggle between perspective projection and orthogonal projection.
   • s: toggle between flat and smooth shading.
   • w: toggle between filled and wireframe mode.
   • t: toggle between static mode and transformation mode: for static mode, the 3D face model doesn't rotate and translate.
   • q: quit the program