DINO_4D: Semantic-Aware 4D Reconstruction

TL;DR

DINO_4D integrates frozen DINOv3 features into 4D scene reconstruction, enhancing semantic awareness and accuracy while maintaining linear time complexity.

Contribution

It introduces a novel method that combines semantic features with geometric reconstruction, improving dynamic scene understanding in real-time.

Findings

Significantly improves Tracking Accuracy (APD) and Reconstruction Completeness.

Maintains linear time complexity $O(T)$ of previous methods.

Establishes a new paradigm for semantic-aware 4D World Models.

Abstract

In the intersection of computer vision and robotic perception, 4D reconstruction of dynamic scenes serve as the critical bridge connecting low-level geometric sensing with high-level semantic understanding. We present DINO\_4D, introducing frozen DINOv3 features as structural priors, injecting semantic awareness into the reconstruction process to effectively suppress semantic drift during dynamic tracking. Experiments on the Point Odyssey and TUM-Dynamics benchmarks demonstrate that our method maintains the linear time complexity $O(T)$ of its predecessors while significantly improving Tracking Accuracy (APD) and Reconstruction Completeness. DINO\_4D establishes a new paradigm for constructing 4D World Models that possess both geometric precision and semantic understanding.