Grand Challenge at ACM MM’26
Estimating the momentary level of engagement from multi-modal participant behaviour is an important prerequisite for assistive systems that support human interactions. At the same time, the relationship between engagement and observable behaviour strongly depends on the social context, cultural background, and characteristics of the interaction. Therefore, MultiMediate'26 poses the challenge of developing engagement estimation approaches that generalise across diverse domains. The challenge combines training and evaluation data from a wide range of interaction scenarios, languages, and participant groups, including adult conversational interactions as well as newly introduced child-child and child-robot play interactions from the PInSoRo dataset, annotated for both social and task engagement. This diversity creates a challenging benchmark for approaches that must transfer across different social situations, languages, age groups, and annotation schemes. In addition to engagement estimation, we continue to invite submissions to popular tasks from previous iterations of the challenge: eye contact detection, bodily behaviour recognition, and backchannel detection.
