Research map / Video / Video segmentation and tracking
BlazeBVD: Make Scale-Time Equalization Great Again for Blind Video Deflickering
Xinmin Qiu, Congying Han, Zicheng Zhang, Bonan Li, Tiande Guo, Pingyu Wang, Xuecheng Nie
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 global flicker removal module (GFRM) (ablation) | BlazeBVD · DAVIS-2017-Test | PSNR: 23.8939 -> 28.6092 | Improved |
| add local flicker removal module (LFRM) (ablation) | BlazeBVD · DAVIS-2017-Test | PSNR: 27.1366 -> 28.6092 | Improved |
| use STE filtering instead of mean filtering in illumination space (ablation) | BlazeBVD · DAVIS-2017-Test | PSNR: 26.4412 -> 28.6092 | Improved |
| add temporal consistency model (TCM) (ablation) | BlazeBVD · DAVIS-2017-Test | E_warp: 0.1093 -> 0.0825 | Improved |
| add TCM (ablation) | BlazeBVD · DAVIS-2017-Test | PSNR: 29.9130 -> 28.6092 | Hurt |
| use adaptive mask weighted warping loss instead of unweighted (ablation) | TCM · DAVIS-2017-Test | E_warp: 0.0849 -> 0.0825 | Improved |
| apply STE in illumination space rather than in color channels separately (ablation) | BlazeBVD · qualitative | color artifacts (qualitative) | Improved |
| replace Deflicker with BlazeBVD | blind video deflickering · 80p video inference | inference time: 614.19s -> 58.37s | Improved |
| replace Deflicker with BlazeBVD | blind video deflickering · Blind Deflickering Dataset synthetic | E_warp: 0.0856 -> 0.0897 | Hurt |
| replace Deflicker with BlazeBVD | blind video deflickering · Blind Deflickering Dataset synthetic | PSNR: 25.3006 -> 28.1464 | Improved |
Closest papers by meaning
- Blind Video Deflickering by Neural Filtering With a Flawed Atlas (CVPR 2023 · similarity 0.88)
- Relightful Video Portrait Harmonization (CVPR 2026 · similarity 0.85)
- BurstDeflicker: A Benchmark Dataset for Flicker Removal in Dynamic Scenes (NeurIPS 2025 · similarity 0.84)
- Parameterized Stripe Attention for Efficient Video Generation (NeurIPS 2026 · similarity 0.84)
- Real-World Efficient Blind Motion Deblurring via Blur Pixel Discretization (CVPR 2024 · similarity 0.84)
- HiCoM: Hierarchical Coherent Motion for Dynamic Streamable Scenes with 3D Gaussian Splatting (NeurIPS 2024 · similarity 0.83)
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.
