Date published:

Emognition dataset

Together with the Emognition team of scientists from the Wroclaw University of Technology, Prof. Lukasz Kaczmarek and Dr. Maciej Behnke have published another psychophysiological database on emotions. The article was published in Scientific Reports.

The Emognition dataset is designed to test methods for recognizing emotions based on physiological reactions and facial expressions. The study's authors collected data from 43 participants who watched short videos evoking nine discrete emotions: amusement, delight, enthusiasm, sympathy, surprise, anger, disgust, fear and sadness. Three wearables were used to record physiological data in parallel with videos of the upper body. After each video clip, participants completed two types of scales: (1) relating to nine discrete emotions and (2) three dimensions of affect: valence, arousal and motivation. The data obtained facilitate the use of different approaches to emotion recognition. Technical validation showed that watching movie clips evokes the desired emotions. The high quality of the signals was also confirmed. The Emognition dataset can be used to answer research questions related to emotion recognition.