The goal of life-long visual learning is to developing techniques that can continuously and autonomously learn from visual data, potentially for years or decades. During this time the system should build an ever-improving base of generic visual information, and use it as background knowledge and context for solving specific computer vision tasks.
In my talk, I will introduce some challenges that one faces when trying to develop life-long learning computer vision systems and I will discuss some of our recent work on attribute-based and other representations that aim at addressing these challenges.
Christoph Lampert is an assistant professor at IST, Austria. He is very well known for prize winning work on using structured output methods to improve the training and performance of sliding windowmethods. He is co-author of one of the ground-breaking papers on using visual attributes to describe and recognize objects. His current interests involve structured learning methods, object detection and recognition, and lifelong visual learning.