Home |
Contact |
Publications |
Code |
Videos |
Projects |
Bio |
Press |
Object Recognition Abstract - Object recognition is a subproblem of the more general problem of perception, and can be defined as follows. Given a scene consisting of one or more objects, can we identify and localize those objects that are sufficiently visible to the sensory system? It is generally assumed that a description of each object to be recognized is available to the computer and can be used to facilitate the task of identification and localization. These descriptions can either be model-based or appearance-based, or a combination of both. Model-based object representation is based on geometric features, whereas appearance-based representation uses a large set of images for training but does not require any insight into the geometric structure of the objects. Object recognition is a key component of many intelligent vision systems, such as those used in hand-eye coordination for bin picking, inspection, and mobile robotics. |
Link |