Reducing the Run-time Complexity of Support Vector Machine Used for Rail Candidates Detection

Support Vector Machine (SVM) is a technique forclassification and regression. It uses a decision surface called hyperplanethat depends on the regularization parameter and training points lying in themargin of the hyperplane. The run-time complexity of SVM may be reduced throughthe hyperplane affected by the regularization parameter. We deal with rails recognition in images taken fromthe camera mounted on the board of the locomotive. For the purpose of railcandidates detection, we deployed an algorithm using SVM. We performed several experimentsunder different settings. In this paper, we introduce an algorithm using SVMand the impact of its regulation parameter as well as others possible onSVM-performance. The main goal is to decrease time-complexity while maintainingclassification success rate.