AI Paper Atlas

Research map / Point clouds / Shape and surface completion

VisTa3D: A Dataset and Benchmark for Vision, Tactile, and 3D Point Clouds-based Thin Object Reconstruction

Shania Guo, Yeongsik Seo, Andrew Fu, Mei Hao, Iris Xia, Jiwon Lee, Xinyi Xie, Hyoungseob Park, Aaron Dollar, Alex Wong

ECCV 2026 · Point clouds › Shape and surface completion · approach: Unsupervised shape correspondence

Read the paperCheck 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
add tactile inputs to depth completion (Tactile-DC vs OmniDC)depth completion · VisTa3D general eval, DefaultA1: 0.834 -> 0.844Improved
add tactile inputs to depth completion (Tactile-DC vs OmniDC)depth completion · VisTa3D object eval, DefaultA1: 0.687 -> 0.719Improved
add tactile inputs to depth completion (Tactile-DC vs OmniDC)depth completion · VisTa3D object eval, Linear FitA1: 0.342 -> 0.487Improved
add tactile inputs to depth completion (Tactile-DC vs OmniDC)depth completion · VisTa3D object eval, Median ScaleA1: 0.439 -> 0.540Improved
add tactile inputs to depth completion (Tactile-DC vs OmniDC)depth completion · VisTa3D object eval, DefaultAbsRel: 0.274 -> 0.205Improved
restrict evaluation to thin-object mask instead of all pixelsOmniDC · VisTa3D, DefaultA1: 0.834 -> 0.687Hurt
restrict evaluation to thin-object mask instead of all pixelsTactile-DC · VisTa3D, DefaultA1: 0.844 -> 0.719Hurt
apply Linear Fit scale alignment instead of raw outputsDepthAnything V2 · VisTa3D object evalA1: 0.000 -> 0.058Improved
apply Linear Fit scale alignment instead of raw outputsOmniDC · VisTa3D object evalA1: 0.687 -> 0.342Hurt
upgrade UniDepth V1 to V2 (after scale correction)UniDepth · VisTa3D object eval, Linear FitA1: 0.049 -> 0.092Improved
upgrade DepthAnything V1 to V2 (after scale correction)DepthAnything · VisTa3D object eval, Median ScaleA1: 0.071 -> 0.087Improved
use multi-view VGGT instead of monocular DepthAnything V2depth estimation · VisTa3D object eval, Median ScaleA1: 0.087 -> 0.221Improved

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