Low-Cost Brain Sensing for Emotion Recognition

This paper presents the first concrete investigation of emotion recognition capability using a low-cost, open-source EEG amplifier entitled OpenBCI. The most important aspect of this type of study is effective emotion elicitation. In existing related state-of-the-art works, movie clips are widely used for audio-visual emotion elicitation. In this study, two-hundred healthy people of various ages participated in an experiment to select effective clips. The methods for selecting the top 60 most effective clips from a total of 120 candidates consist of regular self-assessment, effective tags, and unsupervised learning. An additional 43 participants gathered to view the selected clips to enable the collection of both emotional EEG data and peripheral physiological signals. The data on emotion recognition tasks were analyzed to predict whether the elicited EEG data had a high or low level of valence/arousal, to evaluate the performance of OpenBCI toward emotion-related applications. The experimental results were found to be comparable with those of existing studies from expensive EEG amplifiers. This is because we used a similar emotion elicitation method with the same algorithm for emotion recognition.