3D Vision and Mobile Robotics Research Group

Object Tracking

Object tracking is one of the fundamental tasks in computer vision, with applications ranging from surveillance to augmented reality and mobile robotics. Our lab focuses on simultaneous model creation and tracking in both 2D and 3D settings. For the 2D case, we research the use of deep convolutional features and their use for accurate multi-object tracking and pose estimation.

Our 3D object tracking method is developed specifically for tangible augmented reality (TAR). Most tracking algorithms for augmented reality require a CAD model of the object in order to work. Our research is about developing a method that is able to construct a 3D model of the object immediately before tracking and then uses this model to produce accurate 6 DoF data.