Call for Papers

Important Dates

• Submission Open: January 16, 2026

• Submission deadline: March 02, 2026

• Notification: April 03, 2026

• Camera-ready: April 08, 2026

Topics of interest

Artificial intelligence is advancing rapidly and drawing wide attention. However, bringing visual intelligence into the real world remains a major challenge. Real-world data (RWD) typically exhibits multi-source heterogeneity, scarcity, and imbalance, often with missing values, noise, biases, limited labels, and constraints from complex environments. The AI4RWC workshop takes real-world visual data as its starting point and focuses on building, deploying, monitoring, and continuously improving vision systems in real-world scenarios. We welcome diverse approaches to these challenges, including few-/zero-shot learning, data-centric AI, continual/open-world learning, etc. We also invite knowledge-driven methods, as well as retrieval-augmented generation and large language models (LLMs/VLMs), and agent theory and applications that operate under real-world constraints. The workshop provides an interdisciplinary forum for researchers and practitioners worldwide to share cutting-edge research, innovative methods, and practical experience. Through keynotes, talks, and panels, it connects vision researchers and domain experts to foster collaboration toward trustworthy, effective, and impactful real-world computer vision. Through keynotes, technical presentations, and panel discussions, it will foster interaction between AI researchers and domain experts, cultivate long-term collaborations, and foster collaboration toward trustworthy, effective, and impactful AI in real environments.

We invite submissions of original research papers, case studies, and surveys on AI and applications for interdisciplinary and real-world challenges. Topics of interest include, but are not limited to:

• Few-shot and zero-shot visual learning for real-world imagery 

• Transfer learning and domain adaptation under real-world visual drift 

• Data-centric computer vision for noisy, imbalanced, and heterogeneous datasets 

• Bias detection, concept drift, and robustness evaluation in deployed vision systems 

• Knowledge-enhanced and multimodal vision-language models (VLMs, VLLMs) 

• Retrieval-augmented perception and grounding for vision tasks 

• Vision agents and autonomous visual reasoning in the wild 

• Continual, open-world, and self-supervised visual learning 

• Case studies on deploying vision systems (e.g., healthcare, remote sensing, cultural heritage) 

• Embodied Vision and Perception-Action Coupling: visual navigation, scene understanding, and sensorimotor grounding 

• Simulation-to-Real Transfer: bridging synthetic and real visual domains for embodied intelligence

Instructions for submission 

Archival Track: The workshop welcomes original, novel research visions, applications, and unpublished research papers that are not under review elsewhere. Submissions to this track will be considered for publication in the official CVPR 2026 workshop proceedings. Papers should be between 5 to 8 pages in length for the main text (excluding references), formatted according to the CVPR 2026 paper submission guidelines. There is no page limit for references or appendices. 

Non-Archival Track: We invite submissions of ongoing projects, preliminary results, or work published elsewhere (e.g., on arXiv or at other conferences). Submissions should be up to 4 pages (excl. references), formatted according to the CVPR 2026 paper submission guidelines. These papers will not be included in the official proceedings and can be submitted to future venues. 

All papers should be submitted through the OpenReview Submission Site. Each submission will undergo a rigorous double-blind review by at least two reviewers. 

All accepted papers will be presented as posters during the workshop, and some of them will be selected for short oral presentations. Poster sessions will be conducted onsite with dedicated time for interactive discussions.