Efficient misalignment-robust representation for real-time face recognition



We proposed a novel misalignment-robust representation (MRR) model in order for real-time face recognition. An efficient two-step optimization algorithm with a coarse-to-fine search strategy was developed to implement MRR. MRR has strong robustness to face misalignment coupled with illumination variation and occlusions, and more importantly, it can do face recognition at a real-time speed (less than 1 second under the Matlab programming environment). We evaluated the proposed MRR on various misaligned face recognition and verification tasks.
The extensive experimental results clearly demonstrated that MRR could achieve similar accuracy to state-of-the-arts but with much faster speed, making it a good candidate for use in real-time face recognition systems.

ref :

Yang, Meng, Lei Zhang, and David Zhang. “Efficient misalignment-robust representation for real-time face recognition.” European Conference on Computer Vision. Springer, Berlin, Heidelberg, 2012.

Robust Kernel Representation with Statistical Local Features for Face Recognition

Metaface learning for sparse representation based face recognition


There are no reviews yet.

Be the first to review “Efficient misalignment-robust representation for real-time face recognition”