Deep Face Recognition for Imperfect Human Face Images on Social Media using the CNN Method

Abstract

Face recognition systems nowadays are applied in any gadget. The usual purpose of this technology is to recognize people's information for security purposes on a smartphone. Unfortunately, this face recognition is only available for perfect facial data. In some cases, there are situations where full faces may not be available on social media and recognizing imperfect human facial structure is not a simple task. The aim of this study is to provide the capability to recognize human faces based on partial or imperfect human facial data. This study explores the idea of face recognition based on partial or imperfect human facial data using the state art convolutional neural network-based architecture, especially for the CNN model with the pre-trained VGG-Face model which promises a good recognizing imperfect image of human face results.