AI Paper Atlas

Research map / Image generation / Super-resolution

FinPercep-RM: A Fine-grained Reward Model and Co-evolutionary Curriculum for RL-based Real-world Super-Resolution

Yidi Liu, Zihao Fan, Jie Huang, Jie Xiao, Dong Li, Wenlong Zhang, LEI BAI, Xueyang Fu, Zheng-jun Zha

CVPR 2026 · Image generation › Super-resolution · approach: Efficient lightweight super-resolution · 2 citations (OpenAlex)

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 changedWhereMeasuredEffect
use CLIP-IQA as ReFL rewardDiT4SR · RealLQ250MUSIQ: 71.832 -> 70.456Hurt
use FinPercep-RM as ReFL reward instead of CLIP-IQADiT4SR · RealLQ250MUSIQ: 70.456 -> 73.456Improved
use FinPercep-RM rewardDiT4SR · DrealSRLPIPS: 0.365 -> 0.342Improved
replace CLIP-IQA reward with FinPercep-RM under DPODiT4SR · RealLQ250MUSIQ: 69.234 -> 72.789Improved
replace CLIP-IQA reward with FinPercep-RM under GRPODiT4SR · RealLQ250MUSIQ: 68.567 -> 72.123Improved
add DINO feature-level difference to ground-truth degradation map (ablation)FinPercep-RM · RealLQ250MUSIQ: 72.123 -> 73.456Improved
couple global score with degradation map instead of weighted sum (ablation)FinPercep-RM · RealLQ250MUSIQ: 72.328 -> 73.456Improved
add CCL (parameter-incremental curriculum) (ablation)FinPercep-RM · RealLQ250MUSIQ: 71.982 -> 73.456Improved
use parameter-incremental CCL instead of loss-weight curriculum (ablation)FinPercep-RM · RealLQ250MUSIQ: 72.842 -> 73.456Improved
CLIP-IQA reward on CLIP-IQA metricDiT4SR · DrealSRClipIQA: 0.548 -> 0.562Improved
FinPercep-RM on ClipIQA metric vs CLIP-IQA rewardSeeSR · DrealSRClipIQA: 0.567 -> 0.561No clear effect

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy