RoFace: A robust face representation approach for accurate classification

The recent development of technological applications has made it inevitable to replicate human eyesight talents artificially and the issues requiring particular attention in the ideas of solutions increase in proportion to the number of applications. Facial classification in admittance restriction and video inspection is typically amidst the open-ended applications, where suitable models have been offered to meet users' needs. While it is true that subsequent efforts have led to the proposal of powerful facial recognition models, limiting factors affecting the quality of the results are always considered. These include low-resolution images, partial occlusion of faces and defense against adversarial attacks. The aspect of the input image, the verification of the presence of face occlusion in the image, the motive derived from the image, and the ability to fend off adversarial attacks are all examined by the RoFace formal representation of the face, which is presented in this paper as a solution to these issues. To assess the impact of these components on the classification/recognition accuracy, experiments have been conducted.