Effective Technique for Face Recognition System Using CNN in Attendance Applications

Abstract

Face recognition is a personal identification system that uses individual characteristics of a person to recognize the person's identity. The software which is used for face recognition is MATLAB. The project area is image processing. Different types of ordinal measures are derived from magnitude and phase components of Gabor images. Two orientations and two wavelengths are used. Gabor filters are used in the pre-processing step. Array of Gabor filters is called as filter bank. Convolutional Neural Network (CNN) is used for face recognition, which outperforms the traditional techniques. It has three layers in which Convolution Layer is used for feature extraction. Pooling Layer is used, which crops the particular part of the image containing the object and then resizes it, because classification requires fixed size of an image. SoftMax Layer is used to normalize the un-normalized vectors and make the predictions into the probabilistic forms. Loss function is used to figure out the error. Finally, the face of a student is recognized by using CNN.