Object class recognition using range of multiple computer vision algorithms

Object recognition using computer vision is a process usually split into two phases: object detection and object recognition. This paper focuses on the problem of detecting objects of diverse classes (e.g., faces, text) on images with cluttered scenes. If one know which objects are being targeted, it can apply specialized algorithms for those particular objects; e.g., HAAR classifiers to detect faces, Stroke Width Transform to detect text etc. Our research aims to build a generic object recognition framework, and one of the main layers of this system is the object detection layer. Here we present the specifics of this layer, by focusing on the theoretical background and its practical application. Since this system applies different algorithms (independent from each other) to one image we also consider the possibility of executing these tasks in a parallel fashion.