AI Paper Atlas

Research map / 3D vision / Face and shape

Common Objects in 3D: Large-Scale Learning and Evaluation of Real-Life 3D Category Reconstruction

Jeremy Reizenstein, Roman Shapovalov, Philipp Henzler, Luca Sbordone, Patrick Labatut, David Novotny

ICCV 2021 · 3D vision › Face and shape · approach: Single-view 3D reconstruction · 411 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 Warp-conditioned Embedding (WCE)NeRF · CO3D single-scenePSNR: 23.6 -> 21.0Hurt
add WCESRN · CO3D 10 categories test-unseen vs NerFormerPSNR: 17.6 -> 14.6Hurt
add positional embedding gamma and WCE to SRNSRN · CO3D single-scenePSNR: 15.8 -> 16.9Mixed
add positional embedding gamma to SRN+WCE (ablation)SRN+WCE · CO3D 10 categories train-unseenPSNR: 16.6 -> 17.6Improved
add positional embedding gamma to SRN+WCE (ablation)SRN+WCE · CO3D 10 categories test-unseenPSNR: 14.6 -> 14.4No clear effect
use NerFormer attention instead of WCE-only NeRFNeRF+WCE · CO3D 10 categories test-unseenPSNR: 13.8 -> 17.6Improved
use opacity-based NeRF instead of SDF-based IDRsingle-scene · CO3D single-scenePSNR: 18.5 -> 23.6Improved
use mesh representation (P3DMesh) instead of implicit methodscategory-centric · CO3D 10 categories test-unseenPSNR: 17.6 -> 12.4Hurt
use autodecoder test-time latent optimization instead of NerFormerNeRF+AD · CO3D test-unseenPSNR: 17.6 -> 10.6Hurt
increase number of source views from 1 to 9NerFormer · CO3D 10 categories test-unseenPSNR: 15.1 -> 18.9Improved
use autodecoder SRN+gamma+AD instead of NerFormercategory-centric · CO3D 10 categories train-unseenPSNR: 17.9 -> 21.7Improved
evaluate on all 50 categories rather than 10NerFormer · CO3D test-unseenPSNR: 17.6 -> 15.7Mixed

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