Research map / Image generation / Super-resolution
Adaptive Patch Exiting for Scalable Single Image Super-Resolution
Shizun Wang, Jiaming Liu, Kaixin Chen, Xiaoqi Li, Ming Lu, Yandong Guo
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 adaptive patch exiting with threshold 0 | RCAN · DIV2K x2 SR | PSNR: 34.09dB -> 34.36dB | Improved |
| add adaptive patch exiting with threshold 0 | EDSR · DIV2K x4 SR | PSNR: 28.66dB -> 28.78dB | Improved |
| add adaptive patch exiting with threshold 0 | VDSR · DIV2K x2 SR | PSNR: 33.63dB -> 33.62dB | No clear effect |
| add adaptive patch exiting with threshold 0 | ECBSR · DIV8K x2 SR | PSNR: 39.82dB -> 39.73dB | Hurt |
| add adaptive patch exiting at equal PSNR | RCAN · DIV2K x2 SR | total FLOPs: 100% -> 24% | Improved |
| add adaptive patch exiting at equal PSNR | RCAN · DIV2K x2 SR inference time | time: 2323 -> 974 | Improved |
| add adaptive patch exiting at equal PSNR | EDSR · DIV2K x4 SR | total FLOPs: 100% -> 67% | Improved |
| use exit interval of 4 layers instead of 1 or 2 (ablation) | EDSR-APE · DIV2K x2 SR | performance-efficiency | Improved |
| use incremental capacity instead of absolute performance as exit signal (ablation) | EDSR-APE · DIV2K x2 SR | FLOPs | Improved |
| increase patch size and stride from (32,30) to (48,46) (ablation) | EDSR-APE · DIV2K x4 SR | PSNR: 28.702 -> 28.783 | No clear effect |
| use APE instead of AdaDSR | RCAN · DIV2K x4 SR | inference time: 1644ms -> 350ms | Improved |
| use APE instead of ClassSR | RCAN · DIV2K x4 SR | parameters: 30.1M -> 15.6M | Improved |
Closest papers by meaning
- AdaBM: On-the-Fly Adaptive Bit Mapping for Image Super-Resolution (CVPR 2024 · similarity 0.86)
- Hierarchical Patch Diffusion Models for High-Resolution Video Generation (CVPR 2024 · similarity 0.86)
- Edge-SD-SR: Low Latency and Parameter Efficient On-device Super-Resolution with Stable Diffusion via Bidirectional Conditioning (CVPR 2025 · similarity 0.86)
- AdaFormer: Efficient Transformer with Adaptive Token Sparsification for Image Super-resolution (AAAI 2024 · similarity 0.86)
- SinSR: Diffusion-Based Image Super-Resolution in a Single Step (CVPR 2024 · similarity 0.86)
- ResDiff: Combining CNN and Diffusion Model for Image Super-resolution (AAAI 2024 · 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.
