Student Physical Violence Detection using Convolutional Neural Networks

Abstract

Physical bullying has been an evident issue in the Philippines and Global Context. Thus, it was widely recognized as a major threat in the younger generation in almost every country in the world. It includes wide variety of acts and contexts. As worst reality, physical violence happens mostly in schools. As a result, it affects a person not only physically but also mentally that even leads to death. Nowadays, numerous research in action recognition has been done through Machine Learning to provide an assistance in predicting and recognizing actions using cameras. In this paper, the researcher utilized machine learning algorithms such as Convolutional Neural Networks to train and recognize actions of physical bullying such as Kicking, Punching, and Head Hitting. The findings revealed a positive result in detecting such actions using Convolutional Neural Networks. With such, it enables the prevention of further physical bullying occurrences in the future using CCTV Cameras.