Hybrid CMNV2 model Masked Face Detection Recognition Age and Gender Identification of Photo and Real Time Video Images using Transfer Learning Approach and Deep Learning Techniques

Abstract

Face detection systems have generally been used primarily for non-masked faces, newlinewhich include relevant facial characteristics such as the ears, chin, lips, nose and eyes. newlineMasks are necessary to cover faces in many situations, such as pandemics, crime scenes, newlinemedical settings, high pollution and laboratories. This research objective is concentrated on the upper half of the face, particularly on the features around the eyes, ears,nose and forehead. One of the most challenging factors related to masked face age and gender identification is developing a technique to quickly carry out identification newlineand maintain accuracy without needing people to remove their masks. The challenge of newlineperforming masked facial recognition without asking people to remove their face masks newlinewhile maintaining accuracy is a major problem statement. newline My research aim is to develop and implement a single hybrid model for the face newlinedetection, recognition, age and gender identification of people from the photo and newlinein real-time video images with and without a mask. newline In proposed methodologies, I successfully developed a novel approach by combining newlinea pre-trained modified MobileNetV2+Convolutional Architecture for Fast Feature Embedding (CAFFE), named CMNV2 model for feature extraction,masked face image classification and prediction. newline CAFFE model is used as a face detector and the MobileNetV2 is used for mask newlineidentification. newline The proposed CMNV2 model is effective with combined transfer learning, Deep newlineLearning (DL) and Deep Neural Network (DNN) architecture to extract image features. newline In this research of face detection and classification with and without mask, 5 newlinedifferent layers are added to the pre-trained MobileNetV2. For Masked Face newlineClassification (MFC), Masked Face Age Identification (MFAI) and Masked Face newlineGender Identification (MFGI) a pre-trained MobileNetV2 model with 8 different newlinelayers are added. For MFC and Masked Face Recognition (MFR), 7 different newlinelayers were added to pre-trained MobileNetV2 structure. newlinei newline The artificially masked fac