Behavior change analysis due to violent video gaming using deep learning models

Abstract

In the 21st century, technology has advanced at such a wondrous rate that people resort to this developed medium to carry out most of their tasks, from household chores to daily necessities and almost everything else. People of almost all ages resort to technology to pass their leisure time, and gaming is prevalent among everyone. However, this gaming behavior, whether in online, multiplayer, or single player mode, has significant behavioral changes, both negatively and positively. A significant amount of research was conducted from the early stage of video gaming. After the development of machine learning and deep learning, these techniques were used to predict emotions. Employing a unique approach, a vast amount of YouTube videos were collected from different online gaming streamers and then image and audio datasets comprising hundreds of those videos immersed in these intense gaming sessions were created. By using Facial Expression Recognition (FER) and Speech Emotion Recognition (SER) techniques, an approach was made to find a pattern of behavior change during gaming and over time. For FER, various models were used. Also, different models for SER were used. Some of the best models were used to perform prediction on the image and audio data that we had extracted from the videos. This research contributes significant insights into a player’s emotional change while playing video games.