AI Paper Atlas

Research map / Video / Video segmentation and tracking

Video Object Segmentation-aware Video Frame Interpolation

Jun-Sang Yoo, Hongjae Lee, Seung-Won Jung

ICCV 2023 · Video › Video segmentation and tracking · approach: Matching-based video object segmentation · 6 citations (OpenAlex)

Read the paper

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 VOS-based segmentation and bi-directional consistency losses to VFI trainingAdaCoF · DAVIS 2016J&F: 85.9 -> 87.0Improved
add VOS-VFI trainingAdaCoF · DAVIS 2016PSNR: 25.11 -> 25.03No clear effect
add VOS-VFI trainingAdaCoF, CDFI, IFRNet · DAVIS foreground PSNRforeground PSNR: 25.74 -> 25.80 (CDFI, DAVIS 2016)Improved
add VOS-VFI trainingCDFI · Vimeo90K / UCF101 / Xiph perceptual metricsNIQE, PI, NIMA: PI 3.832 -> 3.822 (Vimeo90K)Improved
add VOS-VFI training to the interpolation modelCDFI · VOT2018 trackingAUC: 58.7 -> 60.1Improved
add VOS-VFI trainingIFRNet · VOT2018AUC: 59.5 -> 60.3Improved
add VOS-VFI trainingCDFI · Linemod pose estimationREP-20px: 63.24 -> 64.09Improved
add segmentation loss (forward and backward) (ablation)CDFI · DAVIS 2016J&F: 86.9 -> 87.2Improved
add bi-directional consistency loss (ablation)CDFI · DAVIS 2016J&F: 86.9 -> 87.5Improved
use both losses together versus either alone (ablation)CDFI · DAVIS 2016J&F: 87.7 vs 87.2 and 87.5Improved
use a single-frame segmentation module instead of a VOS model (ablation)CDFI-VOS · DAVIS 2016J&F: 87.7 -> 86.8Hurt
swap the pretrained VOS model (STCN vs other VOS models) (ablation)CDFI-VOS · DAVIS 2016J&F: 87.7 vs 87.4 and 87.7No clear effect

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 3 of 3 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