Deep Learning for Human Visual Attention Recognition Using Transfer Learning

In this paper, we propose a method for recognizing human visual attention by utilizing the image data. In our method, human visual attention is recognized in three steps: data processing, feature learning, and transfer learning - based classification. We rigorously experimented on the PRIMA dataset from GRAVIR Laboratory, INRIA Rhone-Alpes. The experiment results show that human visual attention can be distinguished by more than 95% accuracy. These achievements have shown the feasibility of real-world human visual attention application such as smart classroom, healthcare, and criminal psychology investigation.