AI Paper Atlas

Research map / 3D vision / NeRF and radiance fields

PeRFception: Perception using Radiance Fields

Yoonwoo Jeong, Seungjoo Shin, Junha Lee, Chris Choy, Anima Anandkumar, Minsu Cho, Jaesik Park

NeurIPS 2022 · 3D vision › NeRF and radiance fields · approach: Sparse-input NeRF regularization · 5 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 background augmentation with probability 0.5 (ablation)ResNet18 · PeRFception-CO3D 2D classificationAcc@1 on PeRFception-CO3D: 81.43 -> 82.05Improved
add background augmentation with probability 0.5 (ablation)ResNet152 · PeRFception-CO3D 2D classificationAcc@1 on PeRFception-CO3D: 84.35 -> 85.28Improved
raise background augmentation probability from 0.5 to 1.0 (ablation)ResNet18 · PeRFception-CO3D 2D classificationAcc@1 on PeRFception-CO3D: 82.05 -> 81.31Hurt
raise background augmentation probability from 0.5 to 1.0 (ablation)ResNet152 · PeRFception-CO3D 2D classificationAcc@1 on PeRFception-CO3D: 85.28 -> 84.06Hurt
train on PeRFception-CO3D (p=0.0) instead of CO3D images, test on PeRFception-CO3DResNet18 · CO3D 2D classificationAcc@1: 80.39 -> 81.43Improved
use ImageNet pretrained weightsResNet50 · CO3D 2D classificationAcc@1 on CO3D: 86.83 -> 89.51Improved
use ImageNet pretrained weights with background augmentation p=0.5ResNet18 · PeRFception-CO3D 2D classificationAcc@1 on PeRFception-CO3D: 82.05 -> 84.97Improved
increase network depth from ResNet18 to ResNet152ResNet · CO3D 2D classificationAcc@1 on CO3D: 84.74 -> 87.75Improved
use density or spherical harmonics as 3D input features instead of ones3D ResNet · PeRFception-CO3D 3D classificationAcc@1/Acc@5Improved
use SH + D as input features for segmentationRes16UNet · PeRFception-ScanNet segmentationmIoU/mAccImproved
vary Plenoxels resolution among 128, 256 and 512 (ablation)Plenoxels · PeRFception-ScanNetAvg. PSNRNo clear effect
use original CO3D point-cloud vs voxel representation for storagePeRFception-CO3D · CO3Dmemory: 1.44TB -> 1.33TBImproved

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