Full-length Papers (Archival, Proceedings Track)
Up to 8 pages, excluding references; double-blind review; accepted papers will appear in the CVPR 2026 Workshop Proceedings.
Computer Vision/CORE Unranked
CVPR 2026
CVPR 2026 Workshop on Emerging Directions in Data for Multimodal Foundation Models
The DataMFM workshop at CVPR 2026 focuses on advancing the data foundations of multimodal AI systems, addressing challenges in data quality, diversity, provenance, and evaluation beyond web-scale collection. It promotes emerging directions such as agentic data generation, world-scale data pipelines, and principled data mixture design to build trustworthy, sustainable multimodal data ecosystems.
Key deadlines
Verified September 9, 2026Full paper
May 20, 2026
AoE
Workshop timeline
Submission and decisions
Full paperKey deadline
May 20, 2026 · AoE
Paper fit
Contribution paths
A strong submission should clearly identify its contribution and evaluate it appropriately.
Short Papers / Extended Abstracts (Non-archival)
Up to 4 pages, excluding references; double-blind review; intended for work-in-progress, datasets, benchmarks, and early-stage ideas.
CVPR 2026 Accepted Papers (Non-archival, Non-anonymous)
Papers accepted to the main CVPR 2026 conference; presented at the workshop but not included in the workshop proceedings.
Research areas in scope
01
Core Research Topics
Data collection, generation, and curation for multimodal foundation modelsData quality improvement, filtering, and pruning for scalable and efficient multimodal trainingData recipes and mixture design for balancing scale, quality, diversity, and coverageSynthetic–real hybrid datasets and multimodal data augmentation for robust model developmentBenchmark renewal, creation, and evaluation design for trustworthy multimodal applicationsDetection and mitigation of dataset contamination in training and evaluationCross-modal alignment and grounding across text, image, audio, and video modalitiesFairness, bias reduction, and inclusive representation in multimodal datasetsData provenance, documentation, licensing, and governance for trustworthy dataset lifecyclesMetrics and frameworks for assessing multimodal data quality, diversity, and contaminationBridging modality gaps between text-rich and vision-centric domainsAgentic synthetic data generation and self-improving data pipelines driven by multimodal or VLA modelsBuilding sustainable, transparent, and community-driven multimodal data ecosystems for next generation foundation models
Policies worth checking twice
- All submissions must use the CVPR 2026 paper template.
- Full-length and short paper submissions undergo double-blind review.
- Submissions must not be simultaneously under review at another conference or journal.
- Accepted papers in the archival track must be presented at the workshop.
- Non-archival submissions (including CVPR 2026 accepted papers) are not included in the workshop proceedings.
- Camera-ready submissions for archival papers are due by April 7, 2026.
Official sources
Compiled from the official call for papers. The organizers’ pages remain authoritative.
Last verified September 9, 2026