AI Paper Atlas

Research map / Video / Video segmentation and tracking

A Flexible Recurrent Residual Pyramid Network for Video Frame Interpolation

Haoxian Zhang, Yang Zhao, Ronggang Wang

ECCV 2020 · Video › Video segmentation and tracking · approach: Video frame interpolation · 41 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
increase pyramid layers from 1 to 4 (ablation)RRPN w/o refinement · Thumos15 (720P)PSNR: 33.56 -> 34.69Improved
increase pyramid layers from 1 to 2 (ablation)RRPN w/o refinement · Our test set (480P)PSNR: 30.74 -> 31.75Improved
increase pyramid layers from 2 to 3 (ablation)RRPN w/o refinement · Our test set (480P)PSNR: 31.75 -> 31.80No clear effect
increase pyramid layers from 1 to 2 (ablation)RRPN w/o refinement · UCF101 (240P)PSNR: 34.64 -> 34.72No clear effect
increase pyramid layers from 2 to 3 (ablation)RRPN w/o refinement · Thumos15 (720P)PSNR: 34.50 -> 34.65Improved
add full refinement network (ablation)RRPN-L2 · Our test set (480P)PSNR: 31.75 -> 31.90Improved
add context maps to refinement network (edge R to whole R) (ablation)RRPN-L2 · Our test set (480P)PSNR: 31.83 -> 31.90Improved
add edge maps to refinement network (context R to whole R) (ablation)RRPN-L2 · Our test set (480P)PSNR: 31.86 -> 31.90Improved
add refinement network (basic R, no context or edge maps) (ablation)RRPN-L2 · Our test set (480P)SSIM: 0.913 -> 0.910Hurt
use more pyramid layers (RRPN-L5) for 4K (ablation)RRPN · H.266 4KPSNR: 34.86 -> 35.28Improved
use four layers instead of three for 1080P (ablation)RRPN · ActivityNet (1080P)PSNR: 30.13 -> 30.08Mixed
remove refinement network to cut costRRPN-L2 · 480P efficiencyruntime (s): 0.12 -> 0.06Improved

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