Comparison of HOG (Histogram of Oriented Gradients) and Haar Cascade Algorithms with A Convolutional Neural Network Based Face Detection Approach
Emine Cengil, Ahmet Çınar · International journal of advance research, ideas and innovations in technology · 2017
Face detection is an important computer vision problem that has been working for years. Security and market research are the areas where face detection is used. Face detection is the first step in some problems such as face recognition, age estimation and face expression detection. Several face detection algorithms have been developed up to now. CNN-based algorithms are the state-of-the art technology in image processing problems, as well as other methods in terms of accuracy rates and speed criteria in face detection problems. In this paper, we propose a face detection algorithm having a model like Alexnet. The method is implemented on images from MUCT and FDDB public datasets. In addition, in the study, the Hog feature descriptor and the methods developed with the haar cascade features are compared with the CNN based method using images of the same dataset. Tests show that the proposed method produces better results than the other two methods.