[CfP] CONTEXTUS Workshop @ ECCV 2026

[Cristina Palmero Cantariño]

[Apologies for cross-posting]

We are pleased to invite submissions for our forthcoming CONTEXTUS Workshop: Understanding Multi-Actor Scene Interaction in Context, to be held in conjunction with ECCV 2026 in Malmö, Sweden.

Real-world environments are inherently interactive and social. Humans rarely act in isolation: behavior unfolds through coordination, negotiation, competition, and collaboration, shaped by shared goals, roles, objects, and evolving contextual constraints. As embodied AI systems increasingly operate alongside people (assisting, collaborating, or communicating), they must move beyond isolated perception toward understanding interaction dynamics, social influence, anticipation, and causal structure in situated, multi-human/agent settings.

The CONTEXTUS workshop addresses a fundamental next step for computer vision: transitioning from object- and action-centric recognition toward socially grounded, context-aware visual intelligence. While major advances have been achieved in detection, tracking, and action recognition, modeling relational dependencies, joint intention, temporal coherence, and how actions influence and are influenced by others within a scene remain as challenging problems. Context, whether spatial, temporal, social, or task-level, is central to meaningful interpretation.

The workshop seeks to bring together researchers working at the intersection of context modeling, multimodal learning, embodied AI, and social intelligence. We encourage submissions presenting novel methods, datasets, benchmarks, theoretical insights, interdisciplinary perspectives, and real-world applications related to socially aware and context-driven scene understanding. Topics of interest include, but are not limited to:

• Multi-human, human-object, and human-scene interaction understanding and generation,
• Context-aware scene understanding and representation learning,
• Context modeling, contextual memory, and long-term temporal reasoning,
• Long-tail and open-vocabulary recognition using contextual and relational cues,
• Context-aware prompting of multimodal foundation models,
• Multimodal retrieval-augmented scene understanding,
• Egocentric and exocentric multimodal perception,
• Human behavior understanding and social signal processing,
• Relational and causal reasoning in videos and multimodal data,
• Knowledge injection, procedural knowledge modeling, and structured human-scene understanding,
• Temporal reasoning, forecasting, and action anticipation,
• Affordance learning and task-oriented scene reasoning,
• Neuro-symbolic learning and structured reasoning approaches,
• Explainability and interpretability in multimodal interaction models,
• Foundation and world models for interaction understanding,
• Embodied AI, human-AI interaction, and behavior-conditioned HCI,
• Domain adaptation, generalization, robustness, and bias mitigation for social scene understanding,
• Datasets and benchmarks for context-aware interaction understanding,
• Applications of the above in robotics, healthcare, education, assistive technologies, and smart environments.

Submission tracks
We welcome submission to three tracks:

1. Novel track: novel work, double-blind peer reviewed.
2. Non-archival track: papers published on or accepted at other venues, to be presented during the workshop as posters to increase visibility and encourage further discussion.
3. Challenge track: novel work, double-blind peer reviewed, related to the ongoing ChaLearn UDIVA-HHOI Challenge associated with the workshop.

Important dates
17 July: Novel track submission deadline
17 July: Non-archival track submission deadline
31 July: Challenge track submission deadline
7 August: Notification of acceptance
13 August: Camera ready
8-9 Sept: CONTEXTUS workshop

Invited speakers
Dima Damen (U. Bristol; Deepmind)
Paul Liang (MIT)
Juan Carlos Niebles (Stanford; Salesforce)

Submission details and more information
Please check the official CONTEXTUS Workshop website: https://lap.chalearn.eu/public/ECCV26-CONTEXTUS

Organizers
Cristina Palmero (King’s College London)
Sergio Escalera (Universitat de Barcelona & Computer Vision Center)
Albert Clapés (Universitat de Barcelona)
Xavier Baró (Universitat de Barcelona)
Daniele Berardini (Istituto Italiano di Tecnologia)
Hugo Jair Escalante (The University of Texas at El Paso & INAOE)
Vittorio Murino (Istituto Italiano di Tecnologia & University of Verona)