Video Object Detection using Variable Threshold

Abstract

Recognition-by-components is a theory of object recognition that accounts for the successful identification of objects despite changes in the size or orientation of the image. RBC explains how moderately degraded images, as well as novel examples of objects, are successfully recognized by the visual system This paper describes a general method for building cascade classifiers from part-based deformable models such as pictorial structures. This paper focuses primarily on the case of star-structured models and show how a simple algorithm based on partial hypothesis pruning can speed up object detection by more than one order of magnitude without sacrificing detection accuracy. It based on two algorithms; the cascade variant dynamic programming algorithm fills values in DP tables and training algorithm for the thresholds used in the cascade.