Emmy Noether Funding for Adam Kortylewski

How can machines start to reliably understand images even in scenarios that are not familiar to them? Dr. Adam Kortylewski and his Emmy Noether group at the Department of Computer Science at the University of Freiburg will be researching this question over the next six years. The German Research Foundation (DFG) is supporting the project with more than 1.7 million euros. The research project on "Robust Computer Vision through Neural Analysis-by-Synthesis with 3D-aware Compositional Network Architectures" will start on July 1, 2022.

Machines with Artificial intelligence (AI) needs to use video cameras and computer vision to perceive our world and to extract information about objects and their properties. So far, computer vision only works reliably when the AI is applied in scenarios that are already very familiar to them. "One of the main problems with computer vision is that current approaches are not reliable in unknown or adverse visibility conditions - for example, when objects are partially occluded, appear in a pose not seen before, or are seen in bad weather," Kortylewski explains. "We plan to overcome these difficulties to make computer vision a reliable component of artificial intelligence, which will benefit a wide range of application areas, such as self-driving cars or autonomous industrial robots."

To that end, the research team will develop computer vision systems that combine techniques from deep learning with computer graphics. These advanced deep neural networks will be able to integrate multiple visual recognition tasks into a joint reasoning process. "With this, we can manage to unify ambiguous perceptions, into a robust, consistent image interpretation," says group leader Kortylewski. Prior to his work at the University of Freiburg, he conducted research on computer vision as a postdoctoral researcher at Johns Hopkins University in the US and at the University of Basel in Switzerland.

The DFG’s financial support for an Emmy Noether AI Group strengthens exceptionally qualified early career researchers dedicated to artificial intelligence methods. The DFG enables career options with a high level of autonomy at an early stage. The funds are initially granted for three years and are available for an extension of another three years.

For more information about Adam Kortylewski

Press image for download
Photo: Bertram Somieski