Scope and Objectives

Multimedia retrieval is entering a transformative era. As we move toward 2026, the field is expanding beyond digital archives into Physical AI, where autonomous systems must reason over multimodal sensory data in real-time. Simultaneously, the rise of generative media and decentralized computing has made Media Verification and Privacy-Preserving Learning critical pillars of trustworthy intelligence. The 7th ICDAR workshop invites researchers to submit "Brave New Ideas" that bridge the gap between digital retrieval, physical interaction, and ethical data governance. We seek contributions that leverage Multimodal AI, Federated Learning, and Neurosymbolic reasoning to make cross-modal analytics more intelligent, secure, and context-aware.

Topics of Interest

We invite original research, system demonstrations, and visionary position papers on topics including, but not limited to:

A. Physical AI & Embodied Multimodal Retrieval

Cross-modal retrieval for robotics and autonomous agents.

Spatio-temporal data analysis for Smart Cities and Digital Twins.

Multimodal sensing for environmental and industrial monitoring (CPS domains).

B. Decentralized & Private Intelligence (Edge/Federated AI)

Federated Learning (FL) for cross-data retrieval without data sharing.

On-device (Edge AI) multimodal processing and modularized AI architectures (e.g., AOP).

Privacy-preserving cross-data analytics and Differential Privacy in retrieval.

C. Media Verification & Content Integrity

Cheapfakes Detection: Identifying out-of-context or subtly manipulated multimedia content.

Multi-modal verification techniques for detecting AI-generated vs. authentic data.

Provenance tracking and watermarking for cross-modal retrieval systems.

D. Advanced Multimodal Communication & Reasoning

Large Multimodal Models (LMMs) for complex cross-data reasoning.

Neurosymbolic AI for explainable and trustworthy multimedia retrieval.

Perceptive AI: Understanding human emotions and context in cross-data streams.

E. Emerging Applications & Diversity

Cross-data analytics for Healthcare, Precision Agriculture, and Sustainable Development (SDGs).

Human-centered AI: Assistive technologies and accessible retrieval interfaces.

Retrieval of "Small Data" vs. "Big Data" in specialized professional domains.

Following the success of ICMR workshops from 2020 to 2025, the 7th ICDAR workshop aims to provide a premier venue for researchers exploring the frontier of cross-modal intelligence. In 2026, the challenge has evolved from simple data alignment to Physical AI—where intelligence interacts with the real world—and Privacy-Preserving Decentralized Learning. Furthermore, the rise of generative media necessitates a focus on Media Verification and Cheapfakes Detection to ensure the integrity of retrieved cross-data. This workshop invites "Brave New Ideas" that leverage Multimodal AI, Federated Learning, and Edge Intelligence to analyze and retrieve cross-data while ensuring ethical data governance.

The accepted papers are expected to be published in the ACM ICMR 2026 proceedings.