The Application of CUDA Architecture in Facial Expression Recognition

This work studies the general method of modern graphics processing unit (GPU) calculating and uses compute unified device architecture (CUDA) to present an implementation of facial expression recognition. We calculate the Gromov-Hausdorff distance between faces, and this measures how far each pair of faces are from being isometric. Finally, The results showed the feasibility of this type of device as a massively parallel data processing tool, and that bigger the scale of face data, better is its performance.