VioDet (Video Violence Detection)

Abstract

In the world we live in today, the sheer volume of video content online and around us, is increasing exponentially. As a matter of fact, Cisco stated that by 2022, videos will make up more than 82% of all consumer internet traffic. With such large volumes of video content being generated, it is a reality that a significant portion of it contains some degree of violence. Numerous psychological studies have shown that exposure to violent video content is a causal risk factor for increased aggressive behavior, aggressive cognition, and aggressive affect and for decreased empathy and prosocial behavior. Many studies can be quoted here, however, the problem is clear. We must detect and stop violent videos. VioDet is the solution. VioDet is a modern AI-based video violence detection mechanism that uses deep learning to automatically detect violent content in videos and consequently, flag them as either violent or non-violent. With the help of this classification, users will have the knowledge about violent videos and can stop their propagation. Another case where VioDet can be helpful is real-time CCTV monitoring where if violence is detected, it can notify the concerned authorities. After selecting a base model that could detect violence in videos accurately, we applied ‘transfer learning’ to get new model that met the requirements of our specific use case. In order to run inference, the model was made accessible through a REST API (built using Flask) and the API was connected to a web application built using React.