| Attention is the cognitive process of selectively concentrating on one thing while ignoring others. As a gateway to later processing steps, efficient and fast attentional allocation is key to rapid processing and understanding of natural scenes and thus predation, escape, and mating - in short, to survival.
Knowing which part of a signal will pass the gatekeeper that we call attention is thus an important aspect for many applications. For example, in robotics, attention models can decrease the computational demands of complex algorithms by focusing them on the potentially most relevant and thus interesting information. In marketing, it is essential that the viewers actually perceive the intended advertisement message. For packaging, it is important that a product package grabs the attention of a shopper, despite the many, well-designed distractors of competitors. Accordingly, early predictions (attention models) and exact knowledge (user studies) of human attention are important assets in such industries.
In this talk, I will focus on some of my recent work on and lessons learned about gaze estimation and attention models at the Karlsruhe Institute of Technology and my eye-tracking spinoff Eyezag. |
