Jan van Gemert is an assistant professor and heads the Computer Vision lab. He is now 2.5 years at the TU-Delft, and his research is on artificial intelligence in the field of ‘Computer Vision’. ‘Computer Vision’ is teaching a computer to recognize images and video. Nowadays this is mainly done through ‘deep learning’ a form of artificial neural networks that allow a computer to learn from unstructured input such as images. Jan specifically does research on how to add knowledge to deep networks to learn more efficiently, requiring less input data.
Jan: “I find it amazing to ask good research questions, doing cutting edge research applied to practical applications. Unbridled creativity and unlimited freedom drew me to computer science. Especially artificial intelligence is fascinating: thinking about how to make a computer think. Visual data is especially captivating as they are so easily interpreted by humans, but are so difficult to understand by a machine. It gives me respect for what the human brain can do.”
This short movie shows an example of a deep learning architecture for pixel wise semantic segmentation from the University of Cambridge.