Pixels 2 Affect

In this line of research, the gameplay footage (screen pixels, and optionally audio) is used as the sole input for predicting the arousal of the player. Deep convolutional neural networks use the frames of the screen as input, and treat the self-annotated arousal traces (provided by the player) as a classification or a preference learning task. The work is innovative as it does not include footage of the player's body or face for prediction, but assumes that the players' affective state is embedded in the gameplay context.