The 2026 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) opens today, June 3, at the Colorado Convention Center in Denver. The flagship computer-vision event runs through June 7 and features a record 4,089 accepted papers selected from 16,092 submissions — a 24% jump in submissions over 2025.
What's New This Year
The program committee says the largest concentrations of accepted work fall in image and video synthesis and generation; vision, language and reasoning; multi-modal learning; 3D from multiview and sensors; and medical and biological vision. Embodied and agentic intelligence — the use of vision models inside robots and autonomous agents — is among the fastest-growing tracks.
Industry Showcase
Nvidia is using CVPR to showcase its Nemotron 3 Nano Omni multimodal model, which combines vision, audio and language into a unified system, alongside new Isaac GR00T research demos following the H2 Plus humanoid robot launch with Unitree. Apple has previewed its CVPR 2026 paper slate on its Machine Learning Research site, with submissions concentrated in 3D scene reconstruction and multimodal perception.
Robotics And Autonomous Driving Front And Center
Tesla and Waymo are both expected to present at workshop sessions covering robotaxi perception and large-scale vision-based driving. The acceptance rate remained competitive at roughly 25%, in line with prior years and consistent with CVPR's reputation as one of the most selective venues in machine learning.
Why CVPR Sets The Agenda
CVPR's program shapes the next year of academic and industry investment in vision research. The 2026 edition lands at a moment when vision-language-action models are increasingly central to robotics and autonomous driving, and where leading labs are publishing both embodied AI breakthroughs and infrastructure work. Conference organisers expect more than 12,000 attendees over the five-day program.
Reporting based on coverage from Apple Machine Learning Research, NVIDIA, PR Newswire and CVPR 2026.