Research map / Video / Video segmentation and tracking
Neural Compression-Based Feature Learning for Video Restoration
Cong Huang, Jiahao Li, Bin Li, Dong Liu, Yan Lu
Check 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 neural compression-based feature learning to the baseline with MV refinement (ablation) | Video restoration model · Set8 σ=50 | PSNR: 29.87 -> 30.29 | Improved |
| Add MV refinement (ablation) | Video restoration baseline · Set8 σ=50 | PSNR: 29.75 -> 29.87 | Improved |
| Add feature attention (ablation) | NCFL-based video restoration model · Set8 σ=50 | PSNR: 30.29 -> 30.45 | Improved |
| Use adaptive quantization instead of no quantization (ablation) | NCFL · Set8 σ=50 | PSNR: 29.98 -> 30.29 | Improved |
| Replace adaptive quantization with a fixed quantization step (ablation) | NCFL · Set8 σ=50 | PSNR: 30.29 -> 29.89 | Hurt |
| Add NCFL (ablation) | Video deraining model · Deraining | PSNR: 27.30 -> 28.11 | Improved |
| Add cross-entropy loss to NCFL (ablation) | Video restoration model · Deraining | PSNR: 27.64 -> 28.11 | Improved |
| Replace MV refinement with normal convolutional layers (ablation) | Video restoration model · Set8 σ=50 | PSNR: 29.87 -> 29.79 | Hurt |
| Replace NCFL with normal convolutional layers (ablation) | Video restoration model · Set8 σ=50 | PSNR: 30.29 -> 29.93 | Hurt |
| Increase cross-entropy loss weight to 1/256 or reduce it to 1/4096 (ablation) | NCFL training · Set8 σ=50 | PSNR: 30.45; 30.25; 30.40 | Mixed |
| Use adaptive quantization rather than no quantization on clean features (ablation) | NCFL · Clean DAVIS features | PSNR: 30.39 -> 30.37 | No clear effect |
| Apply the full method instead of BasicVSR++ | Video denoising · Set8 σ=50 | PSNR: 30.32 -> 30.45 | Improved |
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.
- BasicVSR++ — BasicVSR: The Search for Essential Components in Video Super-Resolution and Beyond (CVPR 2021) · better on 5 of 5 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
- Neural Video Compression with Feature Modulation (CVPR 2024 · similarity 0.87)
- Video Compression With Entropy-Constrained Neural Representations (CVPR 2023 · similarity 0.87)
- Deep Kalman Filtering Network for Video Compression Artifact Reduction (ECCV 2018 · similarity 0.86)
- Learning Efficient Deep Discriminative Spatial and Temporal Networks for Video Deblurring (TPAMI 2025 · similarity 0.85)
- NVRC: Neural Video Representation Compression (NeurIPS 2024 · similarity 0.84)
- Classic Video Denoising in a Machine Learning World: Robust, Fast, and Controllable (CVPR 2025 · similarity 0.84)
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.
