English

Depth Anything 3: Recovering the Visual Space from Any Views

Computer Vision and Pattern Recognition 2025-11-14 v1

Abstract

We present Depth Anything 3 (DA3), a model that predicts spatially consistent geometry from an arbitrary number of visual inputs, with or without known camera poses. In pursuit of minimal modeling, DA3 yields two key insights: a single plain transformer (e.g., vanilla DINO encoder) is sufficient as a backbone without architectural specialization, and a singular depth-ray prediction target obviates the need for complex multi-task learning. Through our teacher-student training paradigm, the model achieves a level of detail and generalization on par with Depth Anything 2 (DA2). We establish a new visual geometry benchmark covering camera pose estimation, any-view geometry and visual rendering. On this benchmark, DA3 sets a new state-of-the-art across all tasks, surpassing prior SOTA VGGT by an average of 44.3% in camera pose accuracy and 25.1% in geometric accuracy. Moreover, it outperforms DA2 in monocular depth estimation. All models are trained exclusively on public academic datasets.

Keywords

Cite

@article{arxiv.2511.10647,
  title  = {Depth Anything 3: Recovering the Visual Space from Any Views},
  author = {Haotong Lin and Sili Chen and Junhao Liew and Donny Y. Chen and Zhenyu Li and Guang Shi and Jiashi Feng and Bingyi Kang},
  journal= {arXiv preprint arXiv:2511.10647},
  year   = {2025}
}

Comments

https://depth-anything-3.github.io/

R2 v1 2026-07-01T07:36:25.055Z