Editor for marking and labeling of object images for binary supervised classification in MATLAB environment

Abstract

The supervised classification of object images requires a labeled training set in the learning phase. The labeled training set is created by a teacher on the base of a set of example images. In certain applications, the example images are not directly available and they have to be created from source images. In such cases, it is convenient to use a specialized editor which allows simultaneous marking and labeling of the example images. This paper brings description of such editor which has been developed for MATLAB. It allows marking of example images of a uniform size in source images and their simultaneous labeling for binary classification. Moreover, it allows generating of labeled training sets of desired features where the list of supported features can be easily extended by users. The editor is described both from user's and programmer's perspective in this paper.