Research map / Video / Video segmentation and tracking
Towards High-Quality and Efficient Video Super-Resolution via Spatial-Temporal Data Overfitting
Gen Li, Jie Ji, Minghai Qin, Wei Niu, Bin Ren, Fatemeh Afghah, Linke Guo, Xiaolong Ma
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 |
|---|---|---|---|
| Group patches across spatial and temporal dimensions by their PSNR and overfit each group separately | WDSR · VSD4K game-15s, ×2 | PSNR: 44.23 -> 45.75 | Improved |
| Group patches across spatial and temporal dimensions by their PSNR and overfit each group separately | WDSR · VSD4K vlog-45s, ×4 | PSNR: 42.87 -> 43.99 | Improved |
| Train one model using only the most informative chunk instead of all spatial-temporal chunks (ablation) | WDSR STDO · vlog-15s, ×2 | PSNR: 50.58 -> 50.42 | Hurt |
| Train one model using only the least informative chunk instead of all spatial-temporal chunks (ablation) | WDSR STDO · vlog-15s, ×2 | PSNR: 50.58 -> 46.89 | Hurt |
| Replace separate chunk-specific models with one data-aware jointly trained model (ablation) | WDSR STDO · VSD4K game-15s, ×2 | PSNR: 45.75 -> 45.46 | Hurt |
| Replace separate chunk-specific models with one data-aware jointly trained model (ablation) | WDSR STDO · VSD4K game-15s, ×3 | PSNR: 40.17 -> 39.87 | Hurt |
| Replace separate chunk-specific models with one data-aware jointly trained model (ablation) | WDSR STDO · VSD4K game-15s, ×4 | PSNR: 38.62 -> 38.14 | Hurt |
| Replace four chunk-specific models with one data-aware jointly trained model (ablation) | WDSR STDO · VSD4K game-15s | number of models: 4 -> 1 | Improved |
| Increase the share of training patches from the most informative chunk (ablation) | JSTDO · Video super-resolution | PSNR | Improved |
| Vary the number of spatial-temporal chunks, with performance best around four (ablation) | STDO · game-15s | PSNR | Mixed |
| Apply STDO to a combined 90-second video containing scene conversions (ablation) | WDSR STDO · game-45s + vlog-45s, ×4 | PSNR: 39.92 dB; 43.99 dB and 37.76 dB | Mixed |
Closest papers by meaning
- Data Overfitting for On-Device Super-Resolution with Dynamic Algorithm and Compiler Co-Design (ECCV 2024 · similarity 0.94)
- Real-Time Single Image and Video Super-Resolution Using an Efficient Sub-Pixel Convolutional Neural Network (CVPR 2016 · similarity 0.89)
- Building an End-to-End Spatial-Temporal Convolutional Network for Video Super-Resolution (AAAI 2017 · similarity 0.88)
- Large Motion Video Super-Resolution with Dual Subnet and Multi-Stage Communicated Upsampling (AAAI 2021 · similarity 0.88)
- SAVSR: Arbitrary-Scale Video Super-Resolution via a Learned Scale-Adaptive Network (AAAI 2024 · similarity 0.87)
- Upscale-A-Video: Temporal-Consistent Diffusion Model for Real-World Video Super-Resolution (CVPR 2024 · 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.
