Imagine: you are going to see a movie with your daughter/son, you are in the line in front of the entrance of the theatre talking to one of your friends, and she enters the room first taking with her the tickets with the seat numbers. When you enter you don’t know where she is. A small embedded system in your coat connected to a few button-sized cameras tells you “She/he is third to the right,” “partly occluded by blond woman.” When you start walking towards her/him, the light is dimmed and the system alerts you “attention handbag on floor,” “attention cane between seats,” “attention popcorn bag on seat.”

 

From an engineering point of view, you will think that the implementation of such a system involves methods from Computer Vision (CV). When you try to join all the pieces, you find that even state-of-the-art CV methods, which are very good at solving restricted problems like object detection (floor, seats), categorization (face, handbag, cane) and identification (daughter/son), are not able to categorize all types of objects in complex scenes nor recognize individual objects like faces when partly occluded, especially with additional complications like different illuminations and viewpoints etc. Just imagine for instance the same scene as above but at an airport lounge or in a disco. Not surprisingly, such a general and flexible system still belongs to science fiction. Nevertheless, we know very well one system that can cope with all such complications—our visual system. So, HV (Human Vision) will provide a solution, or CV based on HV. There’s only one small problem left: we need to know first how HV works.

My principal research interest lies on Human Vision, with special focus on the cortical area, exploring a possible architecture, always taking in mind day-to-day applications. But, everything related with Human or Computer vision is of special interests to me.