Research map / Video / Video segmentation and tracking
Exploring Real-Time Super-Resolution: Benchmarking and Fine-Tuning for Streaming Content
Evgeney Bogatyrev, Khaled Abud, Ivan Molodetskikh, Nikita Alutis, Dmitriy S. Vatolin
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 |
|---|---|---|---|
| fine-tune on StreamSR | RLFN · StreamSR test (2x) | Subjective score: 2.17 -> 2.69 | Improved |
| fine-tune on StreamSR | RLFN · StreamSR test (2x) | PSNR: 37.03 -> 37.63 | Improved |
| fine-tune on StreamSR | SPAN · StreamSR test (2x) | Subjective score: 2.55 -> 3.13 | Improved |
| fine-tune on StreamSR | ESPCN · StreamSR test (2x) | PSNR: 30.71 -> 35.76 | Improved |
| use tanh instead of ReLU activation (ablation) | EfRLFN (ECA) · EfRLFN ablation | SSIM: 0.847 -> 0.865 | Improved |
| use tanh instead of Sigmoid-0.5 (ablation) | EfRLFN (ECA) · EfRLFN ablation | SSIM: 0.856 -> 0.865 | Improved |
| use ECA instead of ESA attention (ablation) | EfRLFN (tanh) · EfRLFN ablation | FPS: 234 -> 314 | Improved |
| use ECA instead of ESA attention (ablation) | EfRLFN (tanh) · EfRLFN ablation | LPIPS: 0.171 -> 0.173 | Hurt |
| use ECA instead of ESA attention (ablation) | EfRLFN (ReLU) · EfRLFN ablation | SSIM: 0.849 -> 0.847 | Hurt |
| use full Charbonnier+VGG+Sobel loss instead of ablated losses (ablation) | EfRLFN · StreamSR 4x | SSIM: 0.865 | Improved |
| train on real LR-HR pairs instead of bicubic synthetic LR (ablation) | EfRLFN · DIV2K | PSNR: 33.983 -> 34.553 | Improved |
| train on real LR-HR pairs instead of bicubic synthetic LR (ablation) | SPAN · DIV2K | PSNR: 32.645 -> 33.511 | Improved |
Closest papers by meaning
- InfVSR: Toward Consistency-Driven Streaming Generative Video Super-Resolution (ICML 2026 · similarity 0.90)
- COMISR: Compression-Informed Video Super-Resolution (ICCV 2021 · similarity 0.89)
- FlashVSR: Towards Real-time Diffusion-Based Streaming Video Super Resolution (CVPR 2026 · similarity 0.89)
- Ultra Flash: Scaling Real-Time Streaming Video Generation to High Resolutions (NeurIPS 2026 · similarity 0.88)
- Arbitrary-Scale Video Super-resolution Guided by Dynamic Context (AAAI 2024 · similarity 0.86)
- VideoGigaGAN: Towards Detail-rich Video Super-Resolution (CVPR 2025 · similarity 0.85)
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.
