NVIDIA Unveils Foundation4D: Unified Feedforward Model for Autonomous Driving Perception and Reconstruction

NVIDIA has introduced Foundation4D, a feedforward unified model that performs geometry estimation, 3D detection, semantic occupancy prediction, and Gaussian reconstruction from multi-camera driving videos, built on a shared DINOv2 backbone with factorized image/view/temporal modeling.

2026-10-23 ~ 2026-10-24 · 2 related posts

• NVIDIA's Foundation4D unifies AV perception and reconstruction, topping Waymo and nuScenes — nvidia · 2026-10-23

1 near-duplicate retellings: zhenjun_zhao