Of Chains and Gains: What Reinforcement Learning changes on Reasoning-based Image Quality Assessment
Reinforcement Learning with Verifiable Rewards (RLVR) has improved the performance of large Vision-Language Models (VLMs) on image quality assessment (IQA). Existing approaches often attribute these gains to improved reasoning before scoring, supported by increasingly plausible Chain-of-Thought (CoT). However, whether...
Yi-Xuan Li, Yuxuan Jiang, Han-Wei Zhu et al.
· 0 citations