EFFECTIVE USE OF ARTIFICIAL INTELLIGENCE FOR THE ENHANCEMENT OF FACIAL RECOGNITION SYSTEM

Abstract

This research paper deals with the development of deep neural vision processing technique for the enhancement of facial recognition and verification. This was motivated by the need to recognize individuals in security-controlled environments. This problem was addressed in this research using methods such as data collection, image acquisition, computer vision, face detection, data pre-processing, data processing, training and face recognition. Mathematical models and universal modeling diagrams were used to design the new system which used discrete wavelet transform model, rendering model, convolutional neural network model to develop an enhanced facial recognition system and implement with Matlab. The system was tested and validated using tenfold cross validation technique and the accuracy achieved was 99.22%. Finally, comparative analysis was performed which compared the performance of the new deep learning algorithm developed and the existing state of the art algorithms; and the result showed a percentage improvement of 0.52% which was very good among other achievements, such as system reliability which was lacking in the conventional system.