Main Article Content
Human detection has become one of the major aspect in the real time modern systems whether it is driver-less vehicles or in disaster management or surveillance. Multiple approaches of machine learning are used to find an efficient and effective way of human detection. The proposed method is mainly applied to address the pose-variant problem of human detection. It reduces the redundancy problem which leads to a slow system. To solve the pose variant and redundancy problem, mutation and crossover concept has been applied over Local Binary Pattern (LBP) and Histogram of Oriented Gradient (HOG) feature set to generate final set . Then combination of feature fusion set of LBP and HOG are fed into Support Vector Machine (SVM) for classification purpose. To improve the performance of detector an unsupervised framework has been used for learning. For post-processing to suppress overlapping and redundant windows - Non-maximal suppression is used . For training and testing purpose, INRIA dataset has been used. The proposed method is compared with HOG, LBP, and HOG-LBP techniques, the result shows that our method outperforms these techniques.
How to Cite
, A. P. V. (2017). Human Detection using Feature Fusion Set of LBP and HOG. International Journal on Future Revolution in Computer Science &Amp; Communication Engineering, 3(9), 261–265. Retrieved from https://ijfrcsce.org/index.php/ijfrcsce/article/view/255