Multimodel approach for pedestrian detection

Good performance of pedestrian detection in an automatic driving system is a necessary task. Many pedestrian detection algorithm use Histogram Oriented Gradient (HOG) for feature extraction and Support Vector Machine (SVM) for classification. Some papers use additional features with HOG, such as Local Binary Pattern (HOG-LBP), to improve the performance. Neural Network and Extreme Learning Machine (ELM) are also used for classification like SVM, but SVM always finds global optimum for classification. This paper adds algorithm to this HOG, SVM system for improving the detection performance. This paper proposes a new method which uses pose of pedestrian. The proposed model outperforms conventional method in SDL dataset.