Humans often experience not just a single basic emotion at a time, but rather a blend of several emotions with varying salience. Despite the importance of such blended emotions, most video-based emotion recognition approaches are designed to recognize single emotions only. The few approaches that have attempted to recognize blended emotions typically cannot assess the relative salience of the emotions within a blend. This limitation largely stems from the lack of datasets containing a substantial number of blended emotion samples annotated with relative salience. To address this shortcoming, we introduce BLEMORE, a novel dataset for multimodal (video, audio) blended emotion recognition that includes information on the relative salience of each emotion within a blend. BLEMORE comprises over 3,000 clips from 58 actors, performing 6 basic emotions and 10 distinct blends, where each blend has 3 different salience configurations (50/50, 70/30, and 30/70). Using this dataset, we conduct extensive evaluations of state-of-the-art video classification approaches on two blended emotion prediction tasks: (1) predicting the presence of emotions in a given sample, and (2) predicting the relative salience of emotions in a blend. Our results show that unimodal classifiers achieve up to 29% presence accuracy and 13% salience accuracy on the validation set, while multimodal methods yield clear improvements, with ImageBind + WavLM reaching 35% presence accuracy and HiCMAE 18% salience accuracy. On the held-out test set, the best models achieve 33% presence accuracy (VideoMAEv2 + HuBERT) and 18% salience accuracy (HiCMAE). In sum, the BLEMORE dataset provides a valuable resource to advancing research on emotion recognition systems that account for the complexity and significance of blended emotion expressions.

Estimating momentary conversational engagement is central to assistive, socially aware AI systems, yet models are typically trained and evaluated within a single domain, limiting real-world robustness. The MultiMediate '25 challenge advances engagement estimation to more challenging, cross-cultural, and multi-domain settings. Building on prior challenge editions, we expand beyond NOXI as the sole training source by introducing NOXI-J, a new multilingual corpus covering Japanese and Chinese interactions, enabling both training and evaluation in diverse linguistic contexts. Although NOXI-J conceptually extends NOXI, we treat it as a distinct domain because linguistic, cultural, capture, and annotation differences induce measurable distribution shifts. In this paper, we present new annotations, precomputed multi-modal features (visual, vocal, and verbal), baseline evaluations, and an analysis of the best performing challenge solutions. Beyond accuracy, we quantify fairness using Conditional Demographic Disparity for gender and language. Our baselines confirm strong in-domain performance (e.g., paralinguistic eGeMAPS and video-transformer features) and reveal notable cross-domain drops, underscoring the challenge of cultural, linguistic, and interactional shifts. Fairness analyses indicate generally small discrepancies for our baselines. We observe the largest disparities for the proposed challenge solutions on the Chinese language test set. All annotations, features, code, and leaderboards are made publicly available to foster sustained progress on robust and fair engagement estimation.

Human emotions are often not expressed directly, but regulated according to internal processes and social display rules. For affective computing systems, an understanding of how users regulate their emotions can be highly useful, for example to provide feedback in job interview training, or in psychotherapeutic scenarios. However, at present no method to automatically classify different emotion regulation strategies in a cross-user scenario exists. At the same time, recent studies showed that instruction-tuned Large Language Models (LLMs) can reach impressive performance across a variety of affect recognition tasks such as categorical emotion recognition or sentiment analysis. While these results are promising, it remains unclear to what extent the representational power of LLMs can be utilized in the more subtle task of classifying users' internal emotion regulation strategy. To close this gap, we make use of the recently introduced Deep corpus for modeling the social display of the emotion shame, where each point in time is annotated with one of seven different emotion regulation classes. We fine-tune Llama2-7B as well as the recently introduced Gemma model using Low-rank Optimization on prompts generated from different sources of information on the Deep corpus. These include verbal and nonverbal behavior, person factors, as well as the results of an indepth interview after the interaction. Our results show, that a fine-tuned Llama2-7B LLM is able to classify the utilized emotion regulation strategy with high accuracy (0.84) without needing access to data from post-interaction interviews. This represents a significant improvement over previous approaches based on Bayesian Networks and highlights the importance of modeling verbal behavior in emotion regulation.

Accurate utterance classification in motivational interviews is crucial to automatically understand the quality and dynamics of client-therapist interaction, and it can serve as a key input for systems mediating such interactions. Motivational interviews exhibit three important characteristics. First, there are two distinct roles, namely client and therapist. Second, they are often highly emotionally charged, which can be expressed both in text and in prosody. Finally, context is of central importance to classify any given utterance. Previous works did not adequately incorporate all of these characteristics into utterance classification approaches for mental health dialogues. In contrast, we present M3TCM, a Multi-modal, Multi-task Context Model for utterance classification. Our approach for the first time employs multi-task learning to effectively model both joint and individual components of therapist and client behaviour. Furthermore, M3TCM integrates information from the text and speech modality as well as the conversation context. With our novel approach, we outperform the state of the art for utterance classification on the recently introduced AnnoMI dataset with a relative improvement of 20% for the client- and by 15% for therapist utterance classification. In extensive ablation studies, we quantify the improvement resulting from each contribution.

Distinguishing target from non-target fixations during visual search is a fundamental building block to understand users’ intended actions and to build effective assistance systems. While prior research indicated the feasibility of classifying target vs. non-target fixations based on eye tracking and electroencephalography (EEG) data, these studies were conducted with explicitly instructed search trajectories, abstract visual stimuli, and disregarded any scene context. This is in stark contrast with the fact that human visual search is largely driven by scene characteristics and raises questions regarding generalizability to more realistic scenarios. To close this gap, we, for the first time, investigate the classification of target vs. non-target fixations during free visual search in realistic scenes. In particular, we conducted a 36-participants user study using a large variety of 140 realistic visual search scenes in two highly relevant application scenarios: searching for icons on desktop backgrounds and finding tools in a cluttered workshop. Our approach based on gaze and EEG features outperforms the previous state-of-the-art approach based on a combination of fixation duration and saccade-related potentials. We perform extensive evaluations to assess the generalizability of our approach across scene types. Our approach significantly advances the ability to distinguish between target and non-target fixations in realistic scenarios, achieving 83.6% accuracy in cross-user evaluations. This substantially outperforms previous methods based on saccade-related potentials, which reached only 56.9% accuracy.

Estimating the momentary level of participant's engagement is an important prerequisite for assistive systems that support human interactions. Previous work has addressed this task in within-domain evaluation scenarios, i.e. training and testing on the same dataset. This is in contrast to real-life scenarios where domain shifts between training and testing data frequently occur. With MultiMediate'24, we present the first challenge addressing multi-domain engagement estimation. As training data, we utilise the NOXI database of dyadic novice-expert interactions. In addition to within-domain test data, we add two new test domains. First, we introduce recordings following the NOXI protocol but covering languages that are not present in the NOXI training data. Second, we collected novel engagement annotations on the MPIIGroupInteraction dataset which consists of group discussions between three to four people. In this way, MultiMediate'24 evaluates the ability of approaches to generalise across factors such as language and cultural background, group size, task, and screen-mediated vs. face-to-face interaction. This paper describes the MultiMediate'24 challenge and presents baseline results. In addition, we discuss selected challenge solutions.

Listeners use short interjections, so-called backchannels, to signify attention or express agreement. The automatic analysis of this behavior is of key importance for human conversation analysis and interactive conversational agents. Current state-of-the-art approaches for backchannel analysis from visual behavior make use of two types of features: features based on body pose and features based on facial behavior. At the same time, transformer neural networks have been established as an effective means to fuse input from different data sources, but they have not yet been applied to backchannel analysis. In this work, we conduct a comprehensive evaluation of multi-modal transformer architectures for automatic backchannel analysis based on pose and facial information. We address both the detection of backchannels as well as the task of estimating the agreement expressed in a backchannel. In evaluations on the MultiMediate'22 backchannel detection challenge, we reach 66.4% accuracy with a one-layer transformer architecture, outperforming the previous state of the art. With a two-layer transformer architecture, we furthermore set a new state of the art (0.0604 MSE) on the task of estimating the amount of agreement expressed in a backchannel.

Automatic analysis of human behaviour is a fundamental prerequisite for the creation of machines that can effectively interact with- and support humans in social interactions. In MultiMediate'23, we address two key human social behaviour analysis tasks for the first time in a controlled challenge: engagement estimation and bodily behaviour recognition in social interactions. This paper describes the MultiMediate'23 challenge and presents novel sets of annotations for both tasks. For engagement estimation we collected novel annotations on the NOvice eXpert Interaction (NOXI) database. For bodily behaviour recognition, we annotated test recordings of the MPIIGroupInteraction corpus with the BBSI annotation scheme. In addition, we present baseline results for both challenge tasks.

Human behavior understanding requires looking at minute details in the large context of a scene containing multiple input modalities. It is necessary as it allows the design of more human-like machines. While transformer approaches have shown great improvements, they face multiple challenges such as lack of data or background noise. To tackle these, we introduce the Forced Attention (FAt) Transformer which utilize forced attention with a modified backbone for input encoding and a use of additional inputs. In addition to improving the performance on different tasks and inputs, the modification requires less time and memory resources. We provide a model for a generalised feature extraction for tasks concerning social signals and behavior analysis. Our focus is on understanding behavior in videos where people are interacting with each other or talking into the camera which simulates the first person point of view in social interaction. FAt Transformers are applied to two downstream tasks: personality recognition and body language recognition. We achieve state-of-the-art results for Udiva v0.5, First Impressions v2 and MPII Group Interaction datasets. We further provide an extensive ablation study of the proposed architecture.

Effective and flexible allocation of visual attention is key for pedestrians who have to navigate to a desired goal under different conditions of urgency and safety preferences. While automatic modelling of pedestrian attention holds great promise to improve simulations of pedestrian behavior, current saliency prediction approaches mostly focus on generic free-viewing scenarios and do not reflect the specific challenges present in pedestrian attention prediction. In this paper, we present Context-SalNET, a novel encoder-decoder architecture that explicitly addresses three key challenges of visual attention prediction in pedestrians: First, Context-SalNET explicitly models the context factors urgency and safety preference in the latent space of the encoder-decoder model. Second, we propose the exponentially weighted mean squared error loss (ew-MSE) that is able to better cope with the fact that only a small part of the ground truth saliency maps consist of non-zero entries. Third, we explicitly model epistemic uncertainty to account for the fact that training data for pedestrian attention prediction is limited. To evaluate Context-SalNET, we recorded the first dataset of pedestrian visual attention in VR that includes explicit variation of the context factors urgency and safety preference. Context-SalNET achieves clear improvements over state-of-the-art saliency prediction approaches as well as over ablations. Our novel dataset will be made fully available and can serve as a valuable resource for further research on pedestrian attention prediction.

Desktop notifications should be noticeable but are also subject to a number of design choices, e.g. concerning their size, placement, or opacity. It is currently unknown, however, how these choices interact with the desktop background and their influence on noticeability. To address this limitation, we introduce a software tool to automatically synthesize realistically looking desktop images for major operating systems and applications. Using these images, we present a user study (N=34) to investigate the noticeability of notifications during a primary task. We are first to show that visual importance of the background at the notification location significantly impacts whether users detect notifications. We analyse the utility of visual importance to compensate for suboptimal design choices with respect to noticeability, e.g. small notification size. Finally, we introduce noticeability maps - 2D maps encoding the predicted noticeability across the desktop and inform designers how to trade-off notification design and noticeability.

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 MPIIGroup Interaction 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.

Body language is an eye-catching social signal and its automatic analysis can significantly advance artificial intelligence systems to understand and actively participate in social interactions. While computer vision has made impressive progress in low-level tasks like head and body pose estimation, the detection of more subtle behaviors such as gesturing, grooming, or fumbling is not well explored. In this paper we present BBSI, the first set of annotations of complex Bodily Behaviors embedded in continuous Social Interactions in a group setting. Based on previous work in psychology, we manually annotated 26 hours of spontaneous human behavior in the MPIIGroupInteraction dataset with 15 distinct body language classes. We present comprehensive descriptive statistics on the resulting dataset as well as results of annotation quality evaluations. For automatic detection of these behaviors, we adapt the Pyramid Dilated Attention Network (PDAN), a state-of-the-art approach for human action detection. We perform experiments using four variants of spatial-temporal features as input to PDAN: Two-Stream Inflated 3D CNN, Temporal Segment Networks, Temporal Shift Module and Swin Transformer. Results are promising and indicate a great room for improvement in this difficult task. Representing a key piece in the puzzle towards automatic understanding of social behavior, BBSI is fully available to the research community.

Emotional expressions are inherently multimodal -- integrating facial behavior, speech, and gaze -- but their automatic recognition is often limited to a single modality, e.g. speech during a phone call. While previous work proposed crossmodal emotion embeddings to improve monomodal recognition performance, despite its importance, an explicit representation of gaze was not included. We propose a new approach to emotion recognition that incorporates an explicit representation of gaze in a crossmodal emotion embedding framework. We show that our method outperforms the previous state of the art for both audio-only and video-only emotion classification on the popular One-Minute Gradual Emotion Recognition dataset. Furthermore, we report extensive ablation experiments and provide detailed insights into the performance of different state-of-the-art gaze representations and integration strategies. Our results not only underline the importance of gaze for emotion recognition but also demonstrate a practical and highly effective approach to leveraging gaze information for this task.

We propose a novel method that leverages human fixations to visually decode the image a person has in mind into a photofit (facial composite). Our method combines three neural networks: An encoder, a scoring network, and a decoder. The encoder extracts image features and predicts a neural activation map for each face looked at by a human observer. A neural scoring network compares the human and neural attention and predicts a relevance score for each extracted image feature. Finally, image features are aggregated into a single feature vector as a linear combination of all features weighted by relevance which a decoder de-codes into the final photofit. We train the neural scoring network on a novel dataset containing gaze data of 19 participants looking at collages of synthetic faces. We show that our method significantly outperforms a mean baseline predictor and report on a human study that shows that we can decode photofits that are visually plausible and close to the observer’s mental image.

Multimodal analysis of group behavior is a key task in human-computer interaction, and in the social and behavioral sciences, but is often limited to more easily controllable laboratory settings or requires elaborate multi-sensor setups and time-consuming manual data annotation. We present ConAn – a usable tool to explore and automatically analyze non-verbal behavior of multiple persons during natural group conversations. In contrast to traditional multi-sensor setups, our tool only requires a single 360° camera and uses state-of-the-art computer vision methods to automatically extract behavioral indicators, such as gaze direction, facial expressions, and speaking activity. As such, our tool allows for easy and fast deployment and supports researchers in understanding individual behavior, group interaction dynamics, and in quantifying user-object interactions. We illustrate the benefits of ConAn on three sample use cases: conversation analysis, assessment of collaboration quality, and impact of technology on audience behavior. Taken together, ConAn represents an important step towards democratizing automatic conversation analysis in HCI and beyond.

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 MPIIGroup Interaction 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

We present the first method to anticipate averted gaze in natural dyadic interactions. The task of anticipating averted gaze, i.e. that a person will not make eye contact in the near future, remains unsolved despite its importance for human social encounters as well as a number of applications, including human-robot interaction or conversational agents. Our multimodal method is based on a long short-term memory (LSTM) network that analyses non-verbal facial cues and speaking behaviour. We empirically evaluate our method for different future time horizons on a novel dataset of 121 YouTube videos of dyadic video conferences (74 hours in total). We investigate person-specific and person-independent performance and demonstrate that our method clearly outperforms baselines in both settings. As such, our work sheds light on the tight interplay between eye contact and other non-verbal signals and underlines the potential of computational modelling and anticipation of averted gaze for interactive applications.

A lack of corpora has so far limited advances in integrating human gaze data as a supervisory signal in neural attention mechanisms for natural language processing (NLP). We propose a novel hybrid text saliency model (TSM) that, for the first time, combines a cognitive model of reading with explicit human gaze supervision in a single machine learning framework. On four different corpora we demonstrate that our hybrid TSM duration predictions are highly correlated with human gaze ground truth. We further propose a novel joint modeling approach to integrate TSM predictions into the attention layer of a network designed for a specific upstream NLP task without the need for any task-specific human gaze data. We demonstrate that our joint model outperforms the state of the art in paraphrase generation on the Quora Question Pairs corpus by more than 10% in BLEU-4 and achieves state of the art performance for sentence compression on the challenging Google Sentence Compression corpus. As such, our work introduces a practical approach for bridging between data-driven and cognitive models and demonstrates a new way to integrate human gaze-guided neural attention into NLP tasks.

Automatic saliency-based recalibration is promising for addressing calibration drift in mobile eye trackers but existing bottom-up saliency methods neglect user's goal-directed visual attention in natural behaviour. By inspecting real-life recordings of egocentric eye tracker cameras, we reveal that users are likely to look at their phones once these appear in view. We propose two novel automatic recalibration methods that exploit mobile phone usage: The first builds saliency maps using the phone location in the egocentric view to identify likely gaze locations. The second uses the occurrence of touch events to recalibrate the eye tracker, thereby enabling privacy-preserving recalibration. Through in-depth evaluations on a recent mobile eye tracking dataset (N=17, 65 hours) we show that our approaches outperform a state-of-the-art saliency approach for automatic recalibration. As such, our approach improves mobile eye tracking and gaze-based interaction, particularly for long-term use.

Automatic detection of emergent leaders in small groups from nonverbal behaviour is a growing research topic in social signal processing but existing methods were evaluated on single datasets – an unrealistic assumption for real-world applications in which systems are required to also work in settings unseen at training time. It therefore remains unclear whether current methods for emergent leadership detection generalise to similar but new settings and to which extent. To overcome this limitation, we are the first to study a cross-dataset evaluation setting for the emergent leadership detection task. We provide evaluations for within- and cross-dataset prediction using two current datasets (PAVIS and MPIIGroupInteraction), as well as an investigation on the robustness of commonly used feature channels and online prediction in the cross-dataset setting. Our evaluations show that using pose and eye contact based features, cross-dataset prediction is possible with an accuracy of 0.68, as such providing another important piece of the puzzle towards real-world emergent leadership detection.

Eye contact is one of the most important non-verbal social cues and fundamental to human interactions. However, detecting eye contact without specialised eye tracking equipment poses significant challenges, particularly for multiple people in real-world settings. We present a novel method to robustly detect eye contact in natural three- and four-person interactions using off-the-shelf ambient cameras. Our method exploits that, during conversations, people tend to look at the person who is currently speaking. Harnessing the correlation between people's gaze and speaking behaviour therefore allows our method to automatically acquire training data during deployment and adaptively train eye contact detectors for each target user. We empirically evaluate the performance of our method on a recent dataset of natural group interactions and demonstrate that it achieves a relative improvement over the state-of-the-art method of more than 60%, and also improves over a head pose based baseline.

Rapport, the close and harmonious relationship in which interaction partners are "in sync" with each other, was shown to result in smoother social interactions, improved collaboration, and improved interpersonal outcomes. In this work, we are first to investigate automatic prediction of low rapport during natural interactions within small groups. This task is challenging given that rapport only manifests in subtle non-verbal signals that are, in addition, subject to influences of group dynamics as well as inter-personal idiosyncrasies. We record videos of unscripted discussions of three to four people using a multi-view camera system and microphones. We analyse a rich set of non-verbal signals for rapport detection, namely facial expressions, hand motion, gaze, speaker turns, and speech prosody. Using facial features, we can detect low rapport with an average precision of 0.7 (chance level at 0.25), while incorporating prior knowledge of participants' personalities can even achieve early prediction without a drop in performance. We further provide a detailed analysis of different feature sets and the amount of information contained in different temporal segments of the interactions.

Visual attention is highly fragmented during mobile interactions, but the erratic nature of attention shifts currently limits attentive user interfaces to adapting after the fact, i.e. after shifts have already happened. We instead study attention forecasting - the challenging task of predicting users' gaze behaviour (overt visual attention) in the near future. We present a novel long-term dataset of everyday mobile phone interactions, continuously recorded from 20 participants engaged in common activities on a university campus over 4.5 hours each (more than 90 hours in total). We propose a proof-of-concept method that uses device-integrated sensors and body-worn cameras to encode rich information on device usage and users' visual scene. We demonstrate that our method can forecast bidirectional attention shifts and predict whether the primary attentional focus is on the handheld mobile device. We study the impact of different feature sets on performance and discuss the significant potential but also remaining challenges of forecasting user attention during mobile interactions.

Previous work on emotion recognition from bodily expressions focused on analysing such expressions in isolation, of individuals or in controlled settings, from a single camera view, or required intrusive motion tracking equipment. We study the problem of emotion recognition from bodily expressions and speech during dyadic (person-person) interactions in a real kitchen instrumented with ambient cameras and microphones. We specifically focus on bodily expressions that are embedded in regular interactions and background activities and recorded without human augmentation to increase naturalness of the expressions. We present a human-validated dataset that contains 224 high-resolution, multi-view video clips and audio recordings of emotionally charged interactions between eight couples of actors. The dataset is fully annotated with categorical labels for four basic emotions (anger, happiness, sadness, and surprise) and continuous labels for valence, activation, power, and anticipation provided by five annotators for each actor. We evaluate vision and audio-based emotion recognition using dense trajectories and a standard audio pipeline and provide insights into the importance of different body parts and audio features for emotion recognition.