Face Recognition Using SVM Fed with Intermediate Output of CNN for Face Detection

We present a face recognition method using support vector machines which utilize intermediate output of convolutional neural networks for face detection. Face detection process carry out the functions both of detecting faces in input image with complex background and of generating the feature vector implicitly representing shape information and spatial arrangement of facial components to supply to support vector machines. Our experiments show that robustness to size variability from 0.8 to 1.2 (relative size in units of area for reference face), demonstrating 100% recognition with 0% false acceptance rate for 600 images of 20 people.