Visual Attention from Dynamic Analysis of Head, Eyes and Salient Objects.

摘要

Face of a person conveys a wealth of information about his/her attentive state. Particularly, head and eyes have the potential to derive where and at what the person is looking. Since humans primarily attend to objects of interest, knowledge of salient objects in the surrounding region can help to accurately infer the focus of visual attention of the person. We present novel computational frameworks and systems to infer visual attention by analyzing dynamics of head, eyes and salient objects. We evaluate proposed systems in intelligent automobile spaces with an emphasis on accurate, robust and continuous performance in the naturalistic driving conditions.