Face detection for football supporters crowd

Face detection and recognition are becoming extremely important and common in a wide range of fields ranging from civil and criminal investigation and surveillance. In addition, the detection and the analysis of supporter behaviors can be of help in football violence investigation in stadiums. The face detection is a better way to summarize and recognize the untoward behaviors in the crowd. Herein, the primary focus of this present article is to propose an effective method of face detection in the crowd of football supporter's. The existing methods perform the detection of the face using the classical face algorithm for detecting faces in the normal scene with some person in it. For the stadium, the number of people in an angle of view of a camera can be hundreds. The variation of the scale of faces can affect the accuracy of detecting all present faces in the scene for which, preprocessing can be used for detecting the faces in the crowd like decomposition of the images in four regions then detect the face in each region. The authors have evaluated few traditional methods on MOA dataset. The obtained results using preprocessing improve face detection in comparison to the existing methods.