Overview and Mission of the Computational Vision Lab

We are constructing automated engineering systems that infer properties of a scene from images and videos.  Such computer vision systems could be useful in automated navigation systems that infer a car’s surroundings from images obtained from a video camera.  Vision systems will aid in internet search for images and video based on visual content rather than relying on annotated text associated with the image or video.  Vision systems may lead to automated diagnosis systems for medicine; for example, systems that study differences in anatomical structures inferred from MRI images of the brain across patients could aid in determining whether a patient has a disease.  In order for vision systems to become ubiquitous, new mathematical models and tools must be developed, and there is a need to understand the fundamental properties of images / videos (e.g., motion, textures, shape, appearance) and its relation to the scene.  We are developing the fundamentals of vision systems by drawing upon diverse areas such as information theory, decision theory, control theory, computational topology, graph theory, partial differential equations, and differential geometry.

Organization and Collaborations

The Computational Vision Lab is directed by Prof. Ganesh Sundaramoorthi and is within the Computer, Electrical, and Mathematical Sciences and Engineering Division, and affiliated with the Geometric Modeling and Scientific Visualization Center of KAUST.

The Computational Vision Lab currently has collaborations within KAUST with the Center for Water Desalination and Reuse, with Saudi Aramco. The Computational Vision Lab has collaborations with the Vision Lab and the Department of Mathematics at the University of California, Los Angeles, the Lab for Computational Computer Vision at the Georgia Institute of Technology, the Scuola Normale Superiore in Pisa, Italy, among other leading universities world-wide.