Computer Vision & Pattern Recognition class project @ Koc University
In this project, we create a real-time video filter that uses face detection, face alignment algorithms, head pose estimation and 3D image rendering with high accuracy. Face detection and face alignment algorithms outputs are used to determine the rotational and translational transform of head pose. We use the cascaded regression model.
Watch Demo
Read Project Report
Root Unity Project is the ‘COMP408/‘ folder.
In the Source Folder run these commands:
Create build folder
$ mkdir build
Enter in to build folder
$ cd build
Enable SSE4 for speed
$ cmake .. -DUSE_SSE4_INSTRUCTIONS=ON
Enable AVX for speed
$ cmake .. -DUSE_AVX_INSTRUCTIONS=ON
Build executables
$ cmake --build . --config Release
In the Source Folder run these commands:
Open unity app:
$ open ./Demo.app/
Wait and run project
$ ./build/project