Research map / Video / Video segmentation and tracking
UPR-Net: A Unified Pyramid Recurrent Network for Video Frame Interpolation
Xin Jin, Longhai Wu, Jie Chen, Youxin Chen, Jayoon Koo, Cheul-Hee Hahm, Zhao-Min Chen
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 changed | Where | Measured | Effect |
|---|---|---|---|
| make flow module recurrent in addition to recurrent synthesis (ablation) | UPR-Net · X-TEST 4K | PSNR: 23.65 -> 30.13 | Improved |
| make synthesis module recurrent in addition to recurrent flow (ablation) | UPR-Net · X-TEST 4K | PSNR: 24.01 -> 30.13 | Improved |
| use recurrent modules for both flow and synthesis (ablation) | UPR-Net · SNU-FILM hard | PSNR: 30.25 -> 30.67 | Improved |
| use recurrent modules for both flow and synthesis (ablation) | UPR-Net · Vimeo90K | PSNR: 35.98 -> 36.03 | No clear effect |
| add iterative synthesis (feed upsampled interpolation) (ablation) | UPR-Net · X-TEST 4K | PSNR: 28.91 -> 30.13 | Improved |
| add iterative synthesis (feed upsampled interpolation) (ablation) | UPR-Net · SNU-FILM hard | PSNR: 30.57 -> 30.67 | Improved |
| add iterative synthesis (feed upsampled interpolation) (ablation) | UPR-Net · Vimeo90K | PSNR: 36.02 -> 36.03 | No clear effect |
| unify flow and synthesis in one pyramid network instead of separate networks (ablation) | UPR-Net · X-TEST 4K | PSNR: 29.89 -> 30.13 | Improved |
| add correlation volume to flow module (ablation) | UPR-Net · Vimeo90K | PSNR: 35.81 -> 36.03 | Improved |
| add correlation volume to flow module (ablation) | UPR-Net · X-TEST 4K | PSNR: 29.51 -> 30.13 | Improved |
| add context features to synthesis (ablation) | UPR-Net · Vimeo90K | PSNR: 35.63 -> 36.03 | Improved |
| add context features to synthesis (ablation) | UPR-Net · X-TEST 4K | PSNR: 29.91 -> 30.13 | Mixed |
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.
- IFRNet large — IFRNet: Intermediate Feature Refine Network for Efficient Frame Interpolation (CVPR 2022) · better on 6 of 6 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
- A Unified Pyramid Recurrent Network for Video Frame Interpolation (CVPR 2023 · similarity 0.99)
- A Flexible Recurrent Residual Pyramid Network for Video Frame Interpolation (ECCV 2020 · similarity 0.91)
- Context-Aware Synthesis for Video Frame Interpolation (CVPR 2018 · similarity 0.91)
- IFRNet: Intermediate Feature Refine Network for Efficient Frame Interpolation (CVPR 2022 · similarity 0.89)
- Disentangled Motion Modeling for Video Frame Interpolation (AAAI 2025 · similarity 0.87)
- EDEN: Enhanced Diffusion for High-quality Large-motion Video Frame Interpolation (CVPR 2025 · similarity 0.87)
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.
