A deep feature based saliency model (DeepFeat) is developed to leverage the understanding of the prediction of human fixations. Conventional saliency models often predict the human visual attention relying on few image cues. Although such models predict fixations on a variety of image complexities, their approaches are limited to the incorporated features. In this project aim, utilize the deep features of convolutional neural networks by combining bottom-up and top-down saliency maps. The proposed framework is applied on deep features of three popular deep convolutional neural networks. We exploit four evaluation metrics to evaluate the correspondence between the proposed framework and the ground truth fixations over two datasets. The key findings of the results demonstrate that the deep features of pre-trained deep convolutional neural networks over the Image Net dataset are strong predictors of the human fixation. The incorporation of bottom-up and top down saliency maps out performs the individual bottom-up and top down implementations. Moreover, in comparison to nine saliency models including four state-of-the-art and five conventional saliency models, our proposed DeepFeat model outperforms the conventional saliency models over all four evaluation metrics. This project is implemented with MATLAB software.