One Scene, Two Depths: Probing Geometric Ambiguity in Monocular Foundation Models

Hugging Face Daily Papers Papers
monocular-depth geometric-ambiguity computer-vision benchmark depth-estimation foundation-models

Summary

Introduces MultiDepth-3k, a benchmark to evaluate depth-layer preferences in monocular depth foundation models, and shows Laplacian Visual Prompting can alter reported depth layers, suggesting complementary geometric hypotheses exist across models.

A faithful 3D world representation should account for layered geometry, where a single camera ray may contain multiple visible and geometrically valid surfaces. Monocular depth estimation, however, reduces this structure to one scalar depth per pixel. Transparent scenes make this ambiguity measurable: the same ray can pass through foreground glass and observe the background, turning the supervised target into a convention of annotation, data, and training rather than a scene-intrinsic truth. A learned predictor exposes this convention as its depth-layer preference. We introduce MultiDepth-3k (MD-3k), a sparse two-layer ordinal benchmark for measuring depth-layer preference and multi-layer spatial relationship accuracy (ML-SRA). On MD-3k, leading depth foundation models exhibit diverse layer preferences under standard RGB input, showing that the same layered geometry can be resolved differently across models. We further find that Laplacian Visual Prompting (LVP), a training-free spectral input transformation, can substantially change the reported layer for certain frozen models. The strongest RGB/LVP pair, DAv2-L, reaches 75.5% ML-SRA. These results suggest that depth foundation models may express complementary geometric hypotheses that standard RGB inference leaves unexpressed. We invite the community to rethink depth supervision and evaluation through an ambiguity-aware lens, where multiple valid 3D interpretations are treated as geometric structure to be measured, preserved, and expressed.
Original Article
View Cached Full Text

Cached at: 06/30/26, 03:37 PM

Paper page - One Scene, Two Depths: Probing Geometric Ambiguity in Monocular Foundation Models

Source: https://huggingface.co/papers/2606.29600

Abstract

Afaithful3Dworldrepresentationshouldaccountforlayeredgeometry,whereasinglecameraraymaycontainmultiplevisibleandgeometricallyvalidsurfaces.Monoculardepthestimation,however,reducesthisstructuretoonescalardepthperpixel.Transparentscenesmakethisambiguitymeasurable:thesameraycanpassthroughforegroundglassandobservethebackground,turningthesupervisedtargetintoaconventionofannotation,data,andtrainingratherthanascene-intrinsictruth.Alearnedpredictorexposesthisconventionasitsdepth-layerpreference.WeintroduceMultiDepth-3k(MD-3k),asparsetwo-layerordinalbenchmarkformeasuringdepth-layerpreferenceandmulti-layerspatialrelationshipaccuracy(ML-SRA).OnMD-3k,leadingdepthfoundationmodelsexhibitdiverselayerpreferencesunderstandardRGBinput,showingthatthesamelayeredgeometrycanberesolveddifferentlyacrossmodels.WefurtherfindthatLaplacianVisualPrompting(LVP),atraining-freespectralinputtransformation,cansubstantiallychangethereportedlayerforcertainfrozenmodels.ThestrongestRGB/LVPpair,DAv2-L,reaches75.5%ML-SRA.TheseresultssuggestthatdepthfoundationmodelsmayexpresscomplementarygeometrichypothesesthatstandardRGBinferenceleavesunexpressed.Weinvitethecommunitytorethinkdepthsupervisionandevaluationthroughanambiguity-awarelens,wheremultiplevalid3Dinterpretationsaretreatedasgeometricstructuretobemeasured,preserved,andexpressed.

View arXiv pageView PDFGitHub13Add to collection

Get this paper in your agent:

hf papers read 2606\.29600

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

Models citing this paper0

No model linking this paper

Cite arxiv.org/abs/2606.29600 in a model README.md to link it from this page.

Datasets citing this paper0

No dataset linking this paper

Cite arxiv.org/abs/2606.29600 in a dataset README.md to link it from this page.

Spaces citing this paper0

No Space linking this paper

Cite arxiv.org/abs/2606.29600 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

Unified Panoramic Geometry Estimation via Multi-View Foundation Models

Hugging Face Daily Papers

PaGeR adapts the multi-view perspective foundation model Depth Anything 3 to predict scale-invariant and metric depth, surface normals, and sky segmentation from a single equirectangular image, using a fixed cubemap representation that keeps VRAM and runtime constant. The paper also releases the ZüriPano and PanoInfinigen datasets.

Geometry Matters: 3D Foundation Priors for Learning Semantic Correspondence

Hugging Face Daily Papers

This paper introduces a post-training framework that leverages 3D priors from SAM3D to improve semantic correspondence in 2D foundation features, addressing issues like left-right confusion and repeated parts. The method uses instance-specific 3D reconstruction without pose annotations or spherical geometry shortcuts.

Unlocking Dense Metric Depth Estimation in VLMs

Hugging Face Daily Papers

DepthVLM enhances Vision-Language Models with a lightweight depth head and unified vision-text supervision, achieving dense metric depth estimation and improved 3D spatial reasoning while maintaining multimodal capabilities.