Development of an attendance registration system based on face recognition technique

Face recognition is one of the biometric technologies commonly used to identify human faces. Face recognition systems have become more popular as a result of a number of benefits, including the fact that facial biological features are immutable and because they are simple to apply. As a result, it is often recognized as the most effective biometric technology in a variety of applications, including distance learning, security, and social networking. However, there is still a long way to go as there are obstacles that need to be overcome to create secure, accurate, and reliable face recognition. In distance learning, facial feature extraction is useful for maintaining face validity since it prevents participants' positions from changing. However, there is a gap between research and practical applications in various fields of face recognition. In addition, the demand for distance learning has increased drastically. This increase is due to various learning obstacles that arise from enforced conditions such as seclusion and social distancing, and the use of face recognition techniques in this area offers further advantages to enhance the learning process. This thesis explores significant problems that arise in face recognition systems, such as pose and illumination variations, and occlusion. To solve these problems, many techniques and algorithms have been proposed. This thesis presents a distance learning registration model based on a new face recognition technique. We have developed two new techniques capable of extracting facial features and addressing the challenges associated with face recognition. The first new model, called multi-descriptor, is based on the well-known method of local binary patterns. It involves many different neighbourhoods of the central pixel. Its unique advantage is that this descriptor allows the use of different neighborhood sizes instead of only one point. This structure ensures reasonable effectiveness and also provides the possibility to obtain a different distribution of features. A face recognition model using the pairwise feature descriptor based on the proposed descriptor was developed in this work, and local binary patterns were created to investigate the similarity and dissimilarity between the two models. For both models, the training was done using the support vector machine method on different face databases to overcome face recognition problems such as camera distance, expression, large head size, and illumination variations. In addition, using deep learning, we present a novel but highly efficient convolutional neural network for improving face recognition, namely In-depth. The technique is based on a combination of sequential and residual identity blocks. This allows us to evaluate the effectiveness of using deeper blocks. The new model has proven to be able to extract features from faces in a highly accurate manner compared to the other state-of-the-art methods. In the distance learning registration process, there are several challenges related to training data limitation, face recognition, and verification. We present a new architecture for face recognition and registration in the case of distance learning. One of the advantages of this model is the features extracted and trained using the proposed model and support vector machine. This immediately lowers the recognition and registration errors. The experiments have shown that our model and the registration model are able to recognize almost all the faces and register the corresponding labels. A new face recognition and registration model makes distance learning more secure, accurate, and reliable, enhancing teaching efficacy and boosting distance learning growth.