The Vision and Learning Group at West Virginia University (WVUVL) is directed by Dr. Gianfranco Doretto and is part of the Lane Department of Computer Science and Electrical Engineering. Our group investigates new and fundamental problems as well as applications in computer vision research. We develop novel techniques for endowing machines with high-level visual processing capabilities, much like humans use their visual system for action. We do so by deriving new machine learning and computational tools that account for the geometry, photometry, materials, semantics, and dynamics of the visual world. The outcome of this research is applicable to areas such as image and video analysis and reasoning, video surveillance, multimedia applications, medical and biomedical imaging applications, biometrics, autonomous navigation, robotics, and many others.
The researchers of WVUVL are interested in studying new methods for interpreting and elaborating the information carried by visual data, such as images and video. Most of the time, such data is the measurement of radiation emitted by a source, after its interaction with the materials of a scene. A typical example is a picture taken by a digital camera that measures the light reflected by the objects in the scene. Another example is an image taken by an MRI machine that measures the radiation of human tissues, stimulated by a magnetic field. The process of understanding visual data is fundamental for building complex systems that can “see”, “interact with”, and “take decisions about” the surrounding world.
Sometimes such vision systems have to be fully autonomous. This is the case when a camera network is exploited for monitoring an area, and revealing the presence of unwanted people in a watch list. It is also the case when an autonomous vehicle endowed with camera sensors is driving in an urban environment towards a destination, with limited to no knowledge of the surroundings, and while interpreting street signs and avoiding obstacles.
Other times vision systems entail the presence of a human in the loop. This is the case when a radiologist applies a computer aided diagnosis system to an imaging modality for a preventive medical check-up. It is also the case when an analyst is querying a video archive, looking for a particular event.
Vision apparatus bring extremely high value to the society, which tomorrow will enjoy improved security in public places with the aid of automated visual surveillance, safer traffic circulation with automated driving assistance, better healthcare with advanced CAD systems, and so on. In order to fully enable such systems, and take them to the next level, it is necessary to tackle fundamental and yet unsolved computer vision problems.