Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Zhang, Jinchang, Kang, Xinrou, Lu, Guoyu
Format: Preprint
Veröffentlicht: 2026
Schlagworte:
Online-Zugang:https://arxiv.org/abs/2603.14702
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866910053995380736
author Zhang, Jinchang
Kang, Xinrou
Lu, Guoyu
author_facet Zhang, Jinchang
Kang, Xinrou
Lu, Guoyu
contents Monocular depth estimation can benefit from autoregressive (AR) generation, but direct AR modeling is hindered by the modality gap between RGB and depth, inefficient pixel-wise generation, and instability in continuous depth prediction. We propose a Fractal Visual Autoregressive Diffusion framework that reformulates depth estimation as a coarse-to-fine, next-scale autoregressive generation process. A VCFR module fuses multi-scale image features with current depth predictions to improve cross-modal conditioning, while a conditional denoising diffusion loss models depth distributions directly in continuous space and mitigates errors caused by discrete quantization. To improve computational efficiency, we organize the scale-wise generators into a fractal recursive architecture, reusing a base visual AR unit in a self-similar hierarchy. We further introduce an uncertainty-aware robust consensus aggregation scheme for multi-sample inference to improve fusion stability and provide a practical pixel-wise reliability estimate. Experiments on standard benchmarks demonstrate strong performance and validate the effectiveness of the proposed design.
format Preprint
id arxiv_https___arxiv_org_abs_2603_14702
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Fractal Autoregressive Depth Estimation with Continuous Token Diffusion
Zhang, Jinchang
Kang, Xinrou
Lu, Guoyu
Computer Vision and Pattern Recognition
Monocular depth estimation can benefit from autoregressive (AR) generation, but direct AR modeling is hindered by the modality gap between RGB and depth, inefficient pixel-wise generation, and instability in continuous depth prediction. We propose a Fractal Visual Autoregressive Diffusion framework that reformulates depth estimation as a coarse-to-fine, next-scale autoregressive generation process. A VCFR module fuses multi-scale image features with current depth predictions to improve cross-modal conditioning, while a conditional denoising diffusion loss models depth distributions directly in continuous space and mitigates errors caused by discrete quantization. To improve computational efficiency, we organize the scale-wise generators into a fractal recursive architecture, reusing a base visual AR unit in a self-similar hierarchy. We further introduce an uncertainty-aware robust consensus aggregation scheme for multi-sample inference to improve fusion stability and provide a practical pixel-wise reliability estimate. Experiments on standard benchmarks demonstrate strong performance and validate the effectiveness of the proposed design.
title Fractal Autoregressive Depth Estimation with Continuous Token Diffusion
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2603.14702