Bio-Inspired Representation Learning for Visual Attention Prediction

Visual Attention Prediction (VAP) is a significant and imperative issue in the field of computer vision. Most existing VAP methods are based on deep learning. However, they do not fully take advantage low-level contrast features while generating visual attention map. In this paper, novel method proposed to generate map via bio-inspired representation The learning combines both high-level semantic simultaneously, which developed by fact that human eye sensitive patches with high objects semantics. composed three main steps: 1) feature extraction, 2) 3) generation. Firstly, extracted from refined VGG16, extraction block network. Secondly, during learning, combined designed densely connected block, concatenate various scale scale. Finally, weighted-fusion layer exploited ultimate obtained representations after Extensive experiments performed demonstrate effectiveness method.