About VITA 2026

Advances in computer vision, multimodal learning, and AR/XR technologies and smart glasses are converging toward Virtual Intelligent Task Assistants (VITAs)—systems that observe, interpret, and guide humans in complex real-world activities. This workshop discusses the most recent research to enable long-term task understanding and assistance. Topics include learning from long streaming videos, egocentric and exocentric video understanding, vision-language models, multimodal reasoning, task/step/procedure prediction, planning and correction, human-AI collaboration and coaching, and new datasets and benchmarks. By fostering dialogue across disciplines, the workshop aims to define the core challenges and opportunities for building practical and generalizable VITAs.