AI Paper Atlas

Research map / 3D vision / Face and shape

In-the-Wild Single Camera 3D Reconstruction Through Moving Water Surfaces

Jinhui Xiong, Wolfgang Heidrich

ICCV 2021 · 3D vision › Face and shape · approach: Neural implicit surface reconstruction · 20 citations (OpenAlex)

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 projection loss (TV on projected heightmap) (ablation)Full model · Synthetic Middlebury (30 frames)avg Euclidean distance: 0.286 -> 0.254Improved
add projection loss (TV on projected heightmap) (ablation)Full model · Synthetic Middlebury (120 frames)avg Euclidean distance: 0.255 -> 0.227Improved
add confidence mask on unreliable rays (ablation)Full model · Synthetic Middlebury (30 frames)avg Euclidean distance: 0.278 -> 0.254Improved
add confidence mask on unreliable rays (ablation)Full model · Synthetic Middlebury (120 frames)avg Euclidean distance: 0.249 -> 0.227Improved
add both confidence mask and projection loss (ablation)Base framework · Synthetic Middlebury (60 frames)avg Euclidean distance: 0.284 -> 0.233Improved
increase number of input frames from 30 to 120Full model · Synthetic Middleburyavg Euclidean distance: 0.254 -> 0.227Improved
jointly optimize water surface and scene instead of decoupling (FSRN surface then triangulation)Lab tank scene · Lab real data (qualitative)reconstruction quality (qualitative)Improved

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