AI Paper Atlas

Research map / 3D vision / NeRF and radiance fields

E2NeRF: Event Enhanced Neural Radiance Fields from Blurry Images

Yunshan Qi, Lin Zhu, Yu Zhang, Jia Li

ICCV 2023 · 3D vision › NeRF and radiance fields · approach: Sparse-input NeRF regularization · 45 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 event rendering loss on top of blur rendering loss (ablation)E2NeRF · Synthetic NeRF scenes (blur+novel avg)PSNR: 27.22 -> 29.67Improved
add blur rendering loss to event-loss-only training (ablation)E2NeRF · Synthetic NeRF scenes (blur+novel avg)PSNR: 22.59 -> 29.67Improved
add event loss (ablation)E2NeRF · Synthetic NeRF scenes, blur viewPSNR: 29.38 -> 29.77Improved
add event loss (ablation)E2NeRF · Synthetic NeRF scenes, novel viewPSNR: 27.98 -> 29.56Improved
add event loss (ablation)E2NeRF · Synthetic NeRF scenes, novel viewLPIPS: .0848 -> .0726Improved
increase number of intermediate poses b beyond 4 (ablation)E2NeRF · lego synthetic scenePSNR/SSIM/LPIPSNo clear effect
decrease event loss weight w from 0.01 toward 1/625 (ablation)E2NeRF · lego synthetic scenePSNR/SSIM/LPIPSMixed
use event-image-based pose estimation instead of COLMAP on blurry imagesE2NeRF · Real-world scenes (30 blurry images)number of poses obtained: 24 -> 30 (letter)Improved
use deblurred images from MPR or D2Net as NeRF inputNeRF · Synthetic blur viewPSNR: MPR-NeRF 27.93 vs MPR 28.12; D2Net-NeRF 27.46 vs D2Net 27.81Mixed
use event data alongside blurry images instead of blurry images only (vs Deblur-NeRF)NeRF · Synthetic novel viewPSNR: 19.93 -> 29.56Improved

Later papers that compared with this method

7 later accepted papers report E2NeRF as a baseline in their results tables; 6 of them report a better number on most of the results they share with it.

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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