Call for Papers

Submission Guidelines

We invite submissions to the Workshop on Agentic AI for Visual Media, which will be held in conjunction with CVPR 202X. This workshop focuses on the emerging intersection of computer vision, multimodal large language models, and agentic AI systems.

The goal is to explore how autonomous, tool-using AI agents can perform complex image and video processing, editing, generation, and evaluation tasks in real-world workflows.

Paper Tracks

Topics of Interest

We invite contributions on (but not limited to):

• Image/Video Processing and Generation — Restoration, enhancement, editing, synthesis, and neural rendering for visual media, including video production, storyboarding, and storytelling.
• Generative AI for Creative Visual Media — Foundation models (GANs, diffusion, multi-modal LLMs) and specialized generative models (fine-tuned models, fine-tuning methods) for content creation, customization, and large-scale deployment.
• Image/Video Understanding and Representation — Captioning, summarization, quality and aesthetic assessment, semantic understanding, alignment with human perception, and the use of LLMs as judges for evaluation and decision support.
• LLM/VLM Post-Training and Adaptation — Knowledge editing and injection, supervised fine-tuning, reinforcement learning based post-training, agent-oriented training strategies, construction and curation of training data, and approaches for learning from raw experience, with a focus on enabling more adaptive and trustworthy multi-modal agents.
• Agentic AI Foundations — Agent architectures, orchestration frameworks, multi-modal reasoning, tool use, and adaptive workflow design.
• Multi-Agent Systems — Collaboration, coordination, communication, and reinforcement learning methods for multi-agent and distributed settings.
• Human–Agent Collaboration — Interactive interfaces, trust, transparency, controllability, and socially responsible design of agentic systems.
• Benchmarks, Datasets, and Evaluation — Standardized datasets, evaluation metrics, and protocols for multi-step, agent-driven workflows, including quality, efficiency, and safety.
• Agent Applications and Deployment — Embodied agents, OS/desktop agents, creative copilots, GUI agents, and integration in media production pipelines.
• Infrastructure and Tooling — Real-time systems, scalable deployment, integration with APIs and existing software ecosystems, and resource-efficient implementations.
• Trust, Safety, and Ethics — Reliability, accountability, fairness, and responsible deployment of agentic AI systems for visual media.

Submission Requirements