Towards Consistent Video Geometry Estimation

Hugging Face Daily Papers Papers

Summary

ViGeo is a transformer-based foundation model that recovers dense and consistent 3D geometry from videos using dynamic chunking attention and a completion-based data refinement framework, achieving state-of-the-art performance across multiple tasks.

This work presents ViGeo, a feed-forward foundation model for recovering spatially dense and temporally consistent geometry from video sequences. Built upon a plain transformer architecture without task-specific architectural modifications, ViGeo supports streaming, full-sequence, and long-video inference within a unified model. The key design is dynamic chunking attention, which exposes the model to both bidirectional and causal temporal contexts during training and allows it to adapt its attention pattern at test time without retraining. To improve supervision quality, we further introduce a completion-based data refinement framework. This framework trains a video depth completion teacher that conditions on sparse and noisy annotations and exploits video/multi-view context to produce dense, temporally coherent, and geometrically reliable training targets. Beyond depth and point maps, ViGeo also predicts surface normals within the same framework. Trained solely on public datasets, ViGeo achieves state-of-the-art performance across online, offline, and long-video depth estimation, surface normal estimation, and video point map estimation.
Original Article
View Cached Full Text

Cached at: 05/29/26, 11:01 AM

Paper page - Towards Consistent Video Geometry Estimation

Source: https://huggingface.co/papers/2605.30060 Published on May 28

·

Submitted byhttps://huggingface.co/pkqbajng

zhuon May 29

Authors:

,

,

,

,

,

,

,

,

,

Abstract

ViGeo is a transformer-based foundation model that recovers dense and consistent 3D geometry from videos using dynamic chunking attention and a completion-based data refinement framework.

This work presents ViGeo, a feed-forward foundation model for recovering spatially dense and temporally consistent geometry from video sequences. Built upon a plaintransformer architecturewithout task-specific architectural modifications, ViGeo supports streaming, full-sequence, and long-video inference within a unified model. The key design isdynamic chunking attention, which exposes the model to both bidirectional and causal temporal contexts during training and allows it to adapt its attention pattern at test time without retraining. To improve supervision quality, we further introduce a completion-based data refinement framework. This framework trains avideo depth completionteacher that conditions on sparse and noisy annotations and exploits video/multi-view contextto produce dense, temporally coherent, and geometrically reliable training targets. Beyond depth and point maps, ViGeo also predictssurface normalswithin the same framework. Trained solely on public datasets, ViGeo achieves state-of-the-art performance across online, offline, and long-videodepth estimation, surface normal estimation, andvideo point map estimation.

View arXiv pageView PDFProject pageGitHub37Add to collection

Get this paper in your agent:

hf papers read 2605\.30060

Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash

Models citing this paper2

#### pkqbajng/VideoLDCM Depth Estimation• Updatedabout 1 hour ago • 1 #### pkqbajng/ViGeo Depth Estimation• Updatedabout 1 hour ago • 1

Datasets citing this paper0

No dataset linking this paper

Cite arxiv.org/abs/2605.30060 in a dataset README.md to link it from this page.

Spaces citing this paper0

No Space linking this paper

Cite arxiv.org/abs/2605.30060 in a Space README.md to link it from this page.

Collections including this paper0

No Collection including this paper

Add this paper to acollectionto link it from this page.

Similar Articles

Quantitative Video World Model Evaluation for Geometric-Consistency

Hugging Face Daily Papers

A quantitative framework called PDI-Bench is introduced for evaluating geometric coherence in generated videos through monocular reconstruction and projective-geometry residuals, revealing geometry-specific failure modes in video generators.

ReViV: Reconstructing the Viewer and the View in 4D from Monocular Egocentric Video

Hugging Face Daily Papers

ReViV is a unified framework for holistic egocentric 4D reconstruction that simultaneously reconstructs viewer (body, hand, gaze) and view (depth, camera trajectory) dynamics from a single monocular RGB video using a Masked Generative Egocentric Transformer, achieving state-of-the-art accuracy and efficiency.