Face Recognition and Skin Classification using SVM and CNN

Face recognition software’s are very helpful to identify a person from the face image database. Most of the times in surveillance cameras partial views of the faces are captured with different angles and resolutions. Thus, it is necessary to efficiently utilize each and every information that is available. There is great demand for alternate biometrics modalities in forensic applications that can handle all kinds of real-word challenges such as partially visible face, pose variation, occlusion in such case palmprint recognition and skin classification can proposed as a potential alternative. Numerous presentation attack instruments like realistic three-dimensional face masks are available in the market that can be used to deceive the face recognition system (FRS). Detecting spoof attack is a challenging issue. The proposed approach for face and palmprint recognition system using support virtual machine (SVM) algorithm along with Local Binary Pattern (LBP) and Histogram of Oriented Gradients (HOG) showed good recognition rate for face images in complex illumination, poses. To detect the presentation attack using silicone masks or other presentation attack instruments skin classification using convolutional neural network (CNN) algorithm is proposed which detects and classifies whether the image is of fake skin or real skin and showed overall accuracy rate of 96%.