Intelligent Violence Video Detection System

Abstract

Due to the busy and stressful lifestyle, humans tend to feel frustrated frequently. This harmful emotional behavior results in violations of several rules, regulations and legislation. Violence is one of the serious issues which emerges due to this situation. It also results in uncontrollable human behavior. This behavior can either be verbal arguments or even physical conflicts. A trend of recording and publishing videos related to these kinds of violations in various platforms can be observed widely at present. Therefore, the terms and conditions of these platforms are subjected to frequent changes. Difficulty in identifying and controlling of violent events will result in an increase of such cases. Due to these reasons, the demand for violence detection systems will be significantly increased. Efficient violent detection systems are lacking currently. But, the usage of artificial intelligence in these systems are further limited. Four major components have been used to achieve this goal. They are video-based, embedded audio-based, abused textbased and thumbnail-based violence detection. The machine learning and image processing techniques are used along with these components to improve the clarity of violence detection.