Deeply Learned Invariant Features for Component-based Facial Recognition

Face recognition underage variation is a challenging problem. It difficult task because ageing an intrinsic variation, not like pose and illumination, which can be controlled. We propose approach to extract invariant features improve facial using components. Can over age progression improved by resizing independently each individual component? The components: eyes, mouth, nose were extracted the Viola-Jones algorithm. Then we utilize eyes region rectangle with upper coordinates detect forehead lower cheeks. proposed work uses Convolutional Neural Network ideal input image size for component according many experiments. sum up scores applying weighted fusion final decision. experiments prove that provides highest score contribution among other ones, cheeks are lowest. conducted on two different databases- MORPH, FG-NET databases. achieves state-of-the-art accuracy reaches 100% dataset results obtained MORPH outperform of related works in literature.