Research map / Video / Video segmentation and tracking
Learned Variational Video Color Propagation
Markus Hofinger, Erich Kobler, Alexander Effland, Thomas Pock
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 |
|---|---|---|---|
| add motion proposals to global proposals (fusion) (ablation) | global-only proposals · DAVIS-2017-val (frame 24) | PSNRab (dB): 31.52 -> 34.20 | Improved |
| add regularization (R) to WM,G (ablation) | LVVCP · DAVIS-2017-val (frame 24) | PSNRab (dB): 34.20 -> 35.01 | Improved |
| add local color proposals (ablation) | WM,G + R · DAVIS-2017-val (frame 24) | PSNRab (dB): 35.01 -> 35.05 | Improved |
| add frame propagation augmentation over 9 frames (ablation) | WM,G,L + R · DAVIS-2017-val (frame 24) | PSNRab (dB): 35.05 -> 35.16 | Improved |
| use fast setting (NG=NL=3) (ablation) | WM,G,L + R + NA=9 · DAVIS-2017-val (frame 24) | PSNRab (dB): 35.16 -> 35.20 | No clear effect |
| add multimodel training (mm) (ablation) | fast LVVCP · DAVIS-2017-val (frame 24) | PSNRab (dB): 35.20 -> 35.23 | Improved |
| use global proposals instead of motion proposals (ablation) | single-candidate variant · DAVIS-2017-val (frame 24) | PSNRab (dB): 29.46 -> 31.52 | Improved |
| fuse motion and global proposals at first frame (ablation) | single-candidate variant · DAVIS-2017-val (frame 1) | PSNRab (dB): 41.00 -> 43.46 | Improved |
| use LVVCP mm fast instead of DVCP | color propagation · NDVCP (frame 45) | PSNRab (dB): 34.60 -> 36.15 | Improved |
| use LVVCP mm fast instead of DVCP | color propagation · NDVCP occluded regions (frame 45) | PSNRab (dB): 32.85 -> 33.98 | Improved |
| use LVVCP mm fast instead of DEB | color propagation · DAVIS-2017-test (frame 45) | PSNRab (dB): 30.45 -> 32.78 | Improved |
| use LVVCP mm fast instead of DVCP | color propagation · NDVCP (frame 45) | LPIPS: 0.0559 -> 0.0455 | Improved |
Closest papers by meaning
- Deep Exemplar-Based Video Colorization (CVPR 2019 · similarity 0.88)
- ColorMNet: A Memory-based Deep Spatial-Temporal Feature Propagation Network for Video Colorization (ECCV 2024 · similarity 0.88)
- Fully Automatic Video Colorization With Self-Regularization and Diversity (CVPR 2019 · similarity 0.88)
- Learning Representations for Automatic Colorization (ECCV 2016 · similarity 0.86)
- Learning Diverse Image Colorization (CVPR 2017 · similarity 0.86)
- Deep Video Prior for Video Consistency and Propagation (TPAMI 2022 · similarity 0.86)
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.
