Man-on-Man Brutality Identification on Video data using Haar Cascade Algorithm
M. G. S. K. Sai Teja, Mukkamalla Ranadeepkumar Reddy, Rajshri Aishwarya · 2020
Fight detection is still an open quandary for video concept detection. There has been little prosperity in engendering an algorithm that can detect the fight in surveillance videos with high-performance values. The proposed methodology is called Haar Cascade. Firstly, motion vector images are trained by using a machine learning algorithm for the consecutive frames. Secondly, motion vector features are extracted to their angles and magnitudes. Afterwards, velocity and angle-based motion co-occurrence features are calculated by considering both current and past motion vectors. The result obtained using the proposed algorithm in different videos got a high rate of success. It discriminates fighting scenes with high true positive rates. By using this methodology, can eradicate violence, crime, fights, etc. This is helpful to reduce violence and make the way to detect fights happen also. While comparing to the existing project, this gives a high rate of detection.