Object detection using scale invariant feature transform

Abstract

An object detection scheme using the Scale Invariant Feature Transform (SIFT) is proposed in this paper. The SIFT extracts distinctive invariant features from images and it is a useful tool for matching between different views of an object. This paper proposes how the SIFT can be used for an object detection problem, especially human detection problem. The Support Vector Machine (SVM) is adopted as the classifier in the proposed scheme. Experiments on INRIA Perdestrian dataset are performed. Preliminary results show that the proposed SIFT-SVM scheme yields promising performance in terms of detection accuracy.