Visual Autoregressive Modelling for Monocular Depth Estimation

Amir El-Ghoussani 1 , André Kaup 1 , Nassir Navab 2 , Gustavo Carneiro 3 and Vasileios Belagiannis 1 1 Friedrich-Alexander University Erlangen-Nuremberg, Germany 2 Technical University of Munich, Germany 3 University of Surrey, United Kingdom [email protected] · 2025

Read Visual Autoregressive Modelling for Monocular Depth Estimation by Amir El-Ghoussani 1 , André Kaup 1 , Nassir Navab 2 , Gustavo Carneiro 3 and Vasileios Belagiannis 1 1 Friedrich-Alexander University Erlangen-Nuremberg, Germany 2 Technical University of Munich, Germany 3 University of Surrey, United Kingdom [email protected] (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2512-22653.txt.

Description

Amir El-Ghoussani, André Kaup, Nassir Navab, Gustavo Carneiro, and Vasileios Belagiannis estimate depth from a single photograph using an autoregressive image model rather than a diffusion model. They adapt a large text-to-image visual autoregressive model, adding a scale-wise conditional upsampling mechanism with classifier-free guidance. Generation then proceeds coarse to fine in ten fixed stages, mirroring how the image model builds pictures scale by scale.

Remarkably, fine-tuning needed only 74,000 synthetic samples. Under that constrained training budget, the method reached state-of-the-art results on indoor benchmarks and stayed strong outdoors. The authors position autoregressive priors as a complementary family of geometry-aware generative models for depth. Advantages include how little data adaptation requires. The recipe also transfers readily to other 3D vision tasks.

Catalog record

CREATORAmir El-Ghoussani 1 , André Kaup 1 , Nassir Navab 2 , Gustavo Carneiro 3 and Vasileios Belagiannis 1 1 Friedrich-Alexander University Erlangen-Nuremberg, Germany 2 Technical University of Munich, Germany 3 University of Surrey, United Kingdom [email protected]
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE33 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2512-22653

Browse the full library · About the Tower of Babel