MultiMediate Grand Challenge 2022

1. 

   MultiMediate’22: Backchannel Detection and Agreement Estimation in Group Interactions

   Philipp Müller, Dominik Schiller, Dominike Thomas, Michael Dietz, Hali Lindsay, Patrick Gebhard, Elisabeth André, Andreas Bulling

   arXiv:2209.09578, pp. 1–6, 2022.

   Abstract Links BibTeX

   Backchannels, i.e. short interjections of the listener, serve important meta-conversational purposes like signifying attention or indicating agreement. Despite their key role, automatic analysis of backchannels in group interactions has been largely neglected so far. The MultiMediate challenge addresses, for the first time, the tasks of backchannel detection and agreement estimation from backchannels in group conversations. This paper describes the MultiMediate challenge and presents a novel set of annotations consisting of 7234 backchannel instances for the MPIIGroupInteraction dataset. Each backchannel was additionally annotated with the extent by which it expresses agreement towards the current speaker. In addition to a an analysis of the collected annotations, we present baseline results for both challenge tasks.

   doi:

   Paper: mueller22_arxiv.pdf

   Paper Access: http://arxiv.org/abs/2209.09578

   @techreport{mueller22_arxiv, title = {MultiMediate'22: Backchannel Detection and Agreement Estimation in Group Interactions}, author = {M{\"{u}}ller, Philipp and Schiller, Dominik and Thomas, Dominike and Dietz, Michael and Lindsay, Hali and Gebhard, Patrick and André, Elisabeth and Bulling, Andreas}, year = {2022}, pages = {1--6}, doi = {}, url = {http://arxiv.org/abs/2209.09578} }

Backchannel Detection

Backchannels serve important meta-conversational purposes like signifying attention or indicating agreement. They can be expressed in a variety of ways - ranging from vocal behaviour (“yes”, “ah-ha”) to subtle nonverbal cues like head nods or hand movements. The backchannel detection sub-challenge focuses on classifying whether a participant of a group interaction expresses a backchannel at a given point in time. Challenge participants will be required to perform this classification based on a 10-second context window of audiovisual recordings of the whole group.

Agreement Estimation

A key function of backchannels is the expression of agreement or disagreement towards the current speaker. It is crucial for artificial mediators to have access to this information to understand the group structure and to intervene to avoid potential escalations. In this sub-challenge, participants will address the task of automatically estimating the amount of agreement expressed in a backchannel. In line with the backchannel detection sub-challenge, a 10-second audiovisual context window containing views on all interactants will be provided.

Organisers

MultiMediate Grand Challenge 2021

1. 

   MultiMediate: Multi-modal Group Behaviour Analysis for Artificial Mediation

   Philipp Müller, Dominik Schiller, Dominike Thomas, Guanhua Zhang, Michael Dietz, Patrick Gebhard, Elisabeth André, Andreas Bulling

   Proc. ACM Multimedia (MM), pp. 4878–4882, 2021.

   Abstract Links BibTeX

   Artificial mediators are promising to support human group conversations but at present their abilities are limited by insufficient progress in group behaviour analysis. The MultiMediate challenge addresses, for the first time, two fundamental group behaviour analysis tasks in well-defined conditions: eye contact detection and next speaker prediction. For training and evaluation, MultiMediate makes use of the MPIIGroupInteraction dataset consisting of 22 three- to four-person discussions as well as of an unpublished test set of six additional discussions. This paper describes the MultiMediate challenge and presents the challenge dataset including novel fine-grained speaking annotations that were collected for the purpose of MultiMediate. Furthermore, we present baseline approaches and ablation studies for both challenge tasks.

   doi: 10.1145/3474085.3479219

   Paper: mueller21_mm.pdf

   @inproceedings{mueller21_mm, title = {MultiMediate: Multi-modal Group Behaviour Analysis for Artificial Mediation}, author = {M{\"{u}}ller, Philipp and Schiller, Dominik and Thomas, Dominike and Zhang, Guanhua and Dietz, Michael and Gebhard, Patrick and André, Elisabeth and Bulling, Andreas}, year = {2021}, pages = {4878--4882}, doi = {10.1145/3474085.3479219}, booktitle = {Proc. ACM Multimedia (MM)} }

Eye Contact Detection Sub-challenge

This sub-challenge focuses on eye contact detection in group interactions from ambient RGB cameras. We define eye contact as a discrete indication of whether a participant is looking at another participants’ face, and if so, who this other participant is. Video and audio recordings over a 10 second context window will be provided as input to provide temporal context for the classification decision. Eye contact has to be detected for the last frame of this context window, making the task formulation also applicable to an online prediction scenario as encountered by artificial mediators.

Next Speaker Prediction Sub-challenge

In the next speaker prediction sub-challenge, approaches need to predict which members of the group will be speaking at a future point in time. Similar to the eye contact detection sub-challenge, video and audio recordings over a 10 second context window will be provided as input. Based on this information, approaches need to predict the speaking status of each participant at one second after the end of the context window.

Evaluation of Participants’ Approaches

For the purpose of this challenge we model the next speaker detection problem as a multi label problem. Hence a model for this task should predict a binary value (speaking = 1, not-speaking = 0) for each participant, for a given sample. As a metric to compare the submitted models we will use the unweighted average recall over all samples (see scikit recall_score(y_true, y_pred, average='macro') function).

For the eye contact detection task the problem is modeled as a multi class problem. Given a specific participant, a submitted model should predict with what other participant he or she is making eye contact. The task is modeled using five classes - one for each participants position (classes 1-4) and an additional class for no eye contact (class 0). To evaluate the performance of this task we will use accuracy as a metric (see scikit accuracy_score(y_true, y_pred) function).

Participants will receive training and validation data that can be used to build solutions for each sub-challenge (eye contact detection and next speaker prediction). The evaluation of these approaches will then be performed remotely on our side with the unpublished test portion of the dataset. For that, participants will create and upload docker images with their solutions that are then evaluated on our systems (for more information regarding the process visit this link).

Organisers