Violence Detection Using Neural Network in real time

Violent Detection has great achievements in developing an surveillance fierce detection systems. Identifying undesirable fight movement was taken care through hand-crafted feature detectors. Occasionally learning models were implemented to do the same. These approaches computed very good accuracies on spot and cinema datasets designed to detect violent sequences. However, these techniques couldn't differentiate and extract unique features from camera motion of the games played dataset. The methodology used in this paper is to understand different violence detection techniques knowing the pros and cones so that an innovative methodology with good accuracy could be designed. A proposal of a framework for violence detection using a Neural Network was made. Compared to other methods of deep learning and machine learning this hybrid approach excels in overall performance.