Advancing Face-to-Face Emotion Communication: A Multimodal Dataset (AFFEC)

Meisam J. Sekiavandi ∗,1,3 , Laurits Dixen ∗,1,3 , Jostein Fimland 1,3 , Sree Keerthi Desu 2 , Antonia-Bianca Zserai 1 , Ye Sul Lee 1 , Maria Barrett 1 , Paolo Burreli 1,3 1 IT University of Copenhagen, 2 Technical University of Denmark 3 Pioneer Centre for Artificial Intelligence, ∗ Equal contribut · 2025

Read Advancing Face-to-Face Emotion Communication: A Multimodal Dataset (AFFEC) by Meisam J. Sekiavandi ∗,1,3 , Laurits Dixen ∗,1,3 , Jostein Fimland 1,3 , Sree Keerthi Desu 2 , Antonia-Bianca Zserai 1 , Ye Sul Lee 1 , Maria Barrett 1 , Paolo Burreli 1,3 1 IT University of Copenhagen, 2 Technical University of Denmark 3 Pioneer Centre for Artificial Intelligence, ∗ Equal contribut (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2504-18969.txt.

Description

Machines that read feelings usually train on posed photos or single signals, missing the back-and-forth of a real conversation. Meisam Sekiavandi and co-authors built a dataset to close that gap, called AFFEC, in this 2025 paper. Seventy-three volunteers acted out eighty-four emotional dialogues across six feelings, yielding thousands of trials and tens of thousands of labels.

Each moment was captured from several angles at once: brain waves, eye movements, skin response, face video, and personality scores. The dataset keeps felt emotion apart from perceived emotion, a split most collections blur. Even simple models beat chance, above all for arousal, and adding personality sharpened the felt-emotion guesses. The authors offer AFFEC for building emotion tools that adapt to the person.

Catalog record

CREATORMeisam J. Sekiavandi ∗,1,3 , Laurits Dixen ∗,1,3 , Jostein Fimland 1,3 , Sree Keerthi Desu 2 , Antonia-Bianca Zserai 1 , Ye Sul Lee 1 , Maria Barrett 1 , Paolo Burreli 1,3 1 IT University of Copenhagen, 2 Technical University of Denmark 3 Pioneer Centre for Artificial Intelligence, ∗ Equal contribut
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE50 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2504-18969

Browse the full library · About the Tower of Babel