A novel fast hybrid face recognition approach using convolutional Kernel extreme learning machine with HOG feature extractor

The ability to identify a person's face from a digitized photo or video frame against a database of faces is known as facial recognition. In the past few years, algorithms that use deep learning to recognize faces have become more popular. The majority of them are predicated on extremely accurate but complicated Convolutional Neural Networks (CNNs), which require a lot of computational power, storage space, and a number of training epochs before they provide satisfying results, and are notably difficult to implement. In an effort to reduce the training time by reducing the number of epochs and increase accuracy, this paper introduces a novel fast hybrid face recognition approach HOG-CKELM, based on CNN that makes use of Kernel based Extreme Learning Machines (KELM) and Histogram of Oriented Gradients (HOG) as facial feature extractor. The effectiveness of the proposed hybrid face recognition technique is evaluated using AT & T, Yale, and JAFFE datasets. When compared to traditional HOG-CNN based techniques, the experimental evaluation indicates that the proposed method for face recognition is capable of achieving excellent performance in terms of accuracy and training time.