AI Paper Atlas

Research map / Video / Video segmentation and tracking

ERDN: Equivalent Receptive Field Deformable Network for Video Deblurring

Bangrui Jiang, Zhihuai Xie, Zhen Xia, Songnan Li, Shan Liu

ECCV 2022 · Video › Video segmentation and tracking · approach: Learned optical flow estimation · 19 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
replace optical flow alignment with deformable convolution (ablation)CDVD-TSP · DVDPSNR: 32.13 -> 32.82Improved
use dilated convolution for equivalent receptive field in offset prediction (ablation)Net-1 · DVDPSNR: 32.82 -> 33.02Improved
use dilated spatial pyramid block instead of single dilated layer (ablation)Net-2 · DVDPSNR: 33.02 -> 33.17Improved
add gradient branch (ablation)Net-3 · DVDPSNR: 33.17 -> 33.31Improved
add gradient branch (ablation)Net-3 · DVDSSIM: 0.9378 -> 0.9395Improved
increase feature pyramid from 1 to 4 layers (ablation)ERDN · DVDPSNR: 31.89 -> 33.31Improved
increase feature pyramid from 4 to 5 layers (ablation)ERDN · DVDPSNR: 33.31 -> 33.20Hurt
use 5 input frames instead of 3ERDN · DVDPSNR: 32.91 -> 33.31Improved
use 5 input frames instead of 3ERDN · GOPROPSNR: 32.20 -> 32.48Improved
use 5 input frames instead of 3ERDN · HFRPSNR: 32.01 -> 32.14Improved

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