The Multimodal Sentiment Analysis Challenge (MuSe) 2022 is dedicated to multimodal sentiment and emotion recognition in different scenarios: https://www.muse-challenge.org

This year, a wide range of different prediction targets is featured:


== Sub-Challenges, Datasets & Features ==

1. Humor Detection Sub-challenge (MuSe-Humor)
Predicting the presence of humor in audio-visual recordings of football press conferences. The challenge is based on the novel Passau Spontaneous Football Coach Humor (Passau-SFCH) dataset which consists of videos of press conferences by 10 different German Bundesliga football coaches and comes with binary labels indicating the presence of humor. In addition to audio and video, transcriptions will be made available.

2. Emotional Reactions Sub-challenge (MuSe-Reaction)
Predicting seven fine-grained emotions (Adoration, Amusement, Anxiety, Disgust, Empathic Pain, Fear, Surprise) in user-generated reactions to emotional stimuli. This challenge utilizes the novel large-scale Hume-React dataset, kindly provided by Hume AI (https://www.hume.ai). Hume-React comprises 75 hours of audiovisual recordings of more than 2000 different subjects.

3. Emotional Stress Sub-challenge (MuSe-Stress)
Predicting the level of emotional valence and psycho-physiological arousal from audio-visual recordings of people in a stress-inducing situation. MuSe-Stress uses the Ulm-TSST data set featuring audio, video, transcriptions and physiological signals (respiratory rate, ECG, BPM). MuSe-Stress is a modified version of last year's task with the same name.

The baseline paper is available here: https://www.researchgate.net/publication/359875358_The_MuSe_2022_Multimodal_Sentiment_Analysis_Challenge_Humor_Emotional_Reactions_and_Stress


== How to Participate ==
Instructions are available at https://www.muse-challenge.org/challenge. Data and features are available upon registration. Links to the baseline model (code, weights) and the preliminary baseline paper are available on the homepage (https://www.muse-challenge.org).


== Organisers ==
Björn W. Schuller (Imperial College London/ audEERING, UK)
Andreas König (University of Passau, GER)
Alan Cowen (Hume AI, USA)
Eva-Maria Meßner (University of Ulm, GER)
Erik Cambria (Nanyang Technological University/ SenticNet, SG)
Shahin Amiriparian (University of Augsburg, GER)
Lukas Christ (University of Augsburg, GER)