Abstract
Human handles the huge amount of information in a complex scene by focusing on a certain portion of the information which is known as visual attention. Human visual system (HVS) is having very strong reasoning compare to current development of automatic computerize systems. Nowadays, computer vision systems stimulate the behavior of human visual attention that has many applications in practice such as objects recognition, tracking, and image cropping. Those systems were designed and implemented to accelerate an automatic processing. Among all those fields, the visual attention domain is a very complicated task to process the objects in real-time. To perform real-time processing, many deep learning techniques have been developed in the past try to simulate the visual attention. In this review article, first we introduce the top-most variants of deep learning algorithms such as convolutional and recurrent neural network models. A state-of-the-art survey is also presented about the advances in the field of employing deep learning models in the field of visual attention. Especially, a comparison is also presented in terms in this paper to show the importance of this field in terms of visual attention. The current and future trends are also described to attract the researchers in this field.