A Self Validation Network for Object-Level Human Attention Estimation

Due to the foveated nature of the human vision system, people can focus their visual attention on only a small region of their visual field at a time, which usually contains a single object.