What is a classifier?
A classifier in the context of computer vision is a definition of a specific visual, rectangular pattern. Computer vision software can use the definition to determine if and where the pattern is present in a digital image. In the image above a mobile app is using a classifier to look for the presence of car in a live feed from the phone camera. Other well-known examples of the use of classifiers is for the detection of human faces in images and video.
Potential uses of classifiers: Object tracking in video and automated content detection in images.
See an example
Or try some
Clondyke Software is working on building an extensive catalog of classifiers for many different types of objects. To see them in action you can visit the Classifier Demo Application:
LBP Classifiers for OpenCV
OpenCV is a free and open source package of software for computer vision. It contains basic tools for producing and using classifiers. Two common types of classifiers supported by OpenCV are Haar and LBP classifiers. LBP means "Local Binary Pattern" (see Wikipedia.org article for further details). One difference between Haar and LBP is that the Haar type operates on floating point numbers while LBP classifiers operate on integers. This has the consequence that LBP classifiers are a lot faster to produce, but this comes at the expense of a little accuracy compared to Haar classifiers.