AI Paper Atlas

Research map / 3D vision / Depth and stereo

Mirror3D: Depth Refinement for Mirror Surfaces

Jiaqi Tan, Weijie Lin, Angel X. Chang, Manolis Savva

CVPR 2021 · 3D vision › Depth and stereo · approach: Foundation models for monocular geometry · 35 citations (OpenAlex)

Check your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
Apply Mirror3DNet to raw sensor depthNYUv2-ref sensor depth · NYUv2-ref test, mirror pixelsRMSE: 1.272 -> 0.766Improved
Apply Mirror3DNet after raw-trained saicsaic · NYUv2-ref test, mirror pixelsRMSE: 1.245 -> 0.749Improved
Apply Mirror3DNet after raw-trained BTSBTS · NYUv2-ref test, mirror pixelsRMSE: 0.903 -> 0.567Improved
Apply Mirror3DNet after raw-trained VNLVNL · NYUv2-ref test, mirror pixelsRMSE: 4.216 -> 3.245Improved
Apply Mirror3DNet after raw-trained saicsaic · NYUv2-raw test, mirror pixelsRMSE: 0.107 -> 0.825Hurt
Use 3 rather than 10 anchor normals (ablation)Mirror3DNet · Matterport3D valSeg-AP: 0.464 -> 0.396Hurt
Use 5 rather than 10 anchor normals (ablation)Mirror3DNet · Matterport3D valSeg-AP: 0.464 -> 0.419Hurt
Use 7 rather than 10 anchor normals (ablation)Mirror3DNet · Matterport3D valSeg-AP: 0.464 -> 0.434Hurt
Use 12 rather than 10 anchor normals (ablation)Mirror3DNet · Matterport3D valSeg-AP: 0.464 -> 0.386Hurt
Use a 15-pixel rather than 25-pixel mirror border (ablation)Mirror3DNet · MP3D-mesh-refRMSE: 0.839 -> 0.851Hurt
Use a 35-pixel rather than 25-pixel mirror border (ablation)Mirror3DNet · MP3D-mesh-refRMSE: 0.839 -> 0.842Hurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy