AI Paper Atlas

Research map / 3D vision / Depth and stereo

Structure from Duplicates: Neural Inverse Graphics from a Pile of Objects

Tianhang Cheng, Wei-Chiu Ma, Kaiyu Guan, Antonio Torralba, Shenlong Wang

NeurIPS 2023 · 3D vision › Depth and stereo · approach: Feed-forward 3D reconstruction

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
use duplicate-object sharing instead of InvRender (multi-view)inverse rendering · Multi-view syntheticalbedo PSNR: 16.984 -> 21.961Improved
use duplicate-object sharing instead of InvRender (multi-view)inverse rendering · Multi-view syntheticgeometry Chamfer distance: 0.024 -> 0.011Improved
train on single view versus baselines on 10 viewsinverse rendering · Single-view syntheticalbedo PSNR: 14.977 -> 17.629Improved
train on single view versus baselines on 10 viewsinverse rendering · Single-view syntheticgeometry Chamfer distance: 0.033 -> 0.034No clear effect
single view versus multi-view baselines on real datainverse rendering · Real-world single-viewrendering PSNR: 20.665 -> 20.326No clear effect
single view versus multi-view baselines on real datainverse rendering · Real-world single-viewLPIPS: 0.262 -> 0.192Improved
remove mask loss (ablation)full model · Loss ablationalbedo PSNR: 17.27 -> 15.14Hurt
remove pretrained surface normal prior (ablation)full model · Loss ablationalbedo PSNR: 17.27 -> 16.51Hurt
remove Eikonal loss (ablation)full model · Loss ablationrelighting PSNR: 19.95 -> 19.31Hurt
remove metallic loss (ablation)full model · Loss ablationmetallic MSE: 0.02 -> 0.08Hurt
remove latent smooth regularization (ablation)full model · Loss ablationrelighting PSNR: 19.95 -> 19.71Hurt
increase number of duplicates from 6 to 30full model · color box datasetalbedo PSNR: 18.48 -> 19.69Improved

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