Research map / Architectures and efficiency / Quantization
PAMS: Quantized Super-Resolution via Parameterized Max Scale
Huixia Li, Chenqian Yan, Shaohui Lin, Xiawu Zheng, Baochang Zhang, Fan Yang, Rongrong Ji
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 EMA-based initialization of alpha instead of random (ablation) | PAMS-EDSR 4-bit · Urban100 x4 | PSNR (dB): 23.488 -> 25.806 | Improved |
| add EMA-based initialization of alpha instead of random (ablation) | PAMS-EDSR · Set5 x4 | PSNR (dB): 31.782 -> 32.002 | Improved |
| add SKT loss (ablation) | PAMS-EDSR 4-bit · Urban100 x4 | PSNR (dB): 25.250 -> 25.321 | Improved |
| add SKT loss (ablation) | PAMS-EDSR 8-bit · Urban100 x4 | PSNR (dB): 25.984 -> 26.016 | Improved |
| add SKT loss (ablation) | PAMS-EDSR 8-bit · Set5 x4 | PSNR (dB): 32.127 -> 32.124 | No clear effect |
| replace fixed max scale (TF Lite) with learnable alpha (ablation) | EDSR 8-bit · Urban100 x4 | PSNR (dB): 25.739 -> 26.016 | Improved |
| replace Dorefa quantization with PAMS | EDSR 8-bit · Set14 x4 | PSNR (dB): 27.297 -> 28.585 | Improved |
| replace PACT with PAMS | EDSR 4-bit · Urban100 x4 | PSNR (dB): 25.148 -> 25.321 | Improved |
| use PAMS without BN instead of PACT without BN | EDSR 8-bit · Urban100 x4 | PSNR gap to full precision (dB): 0.790 -> -0.019 | Improved |
| remove BN layers from EDSR under PACT quantization (ablation) | PACT-EDSR 8-bit · Urban100 x4 | PSNR gap to full precision (dB): 0.354 -> 0.790 | Hurt |
| quantize to 4-bit instead of 8-bit | PAMS-RDN · Urban100 x2 | PSNR (dB): 32.262 -> 29.703 | Hurt |
| quantize to 4-bit instead of 8-bit | PAMS-EDSR · BSD100 x4 | PSNR (dB): 27.565 -> 27.322 | Hurt |
Closest papers by meaning
- QuantSR: Accurate Low-bit Quantization for Efficient Image Super-Resolution (NeurIPS 2023 · similarity 0.89)
- QuantSR+: Pushing the Limit of Quantized Image Super-Resolution Networks (TPAMI 2026 · similarity 0.89)
- Gradient Knows Best: Mixed-Precision Quantization via Gradient-Guided Bit Allocation for Super-Resolution (CVPR 2026 · similarity 0.88)
- 2DQuant: Low-bit Post-Training Quantization for Image Super-Resolution (NeurIPS 2024 · similarity 0.86)
- SAVSR: Arbitrary-Scale Video Super-Resolution via a Learned Scale-Adaptive Network (AAAI 2024 · similarity 0.85)
- PassionSR: Post-Training Quantization with Adaptive Scale in One-Step Diffusion based Image 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.
