An improved mechanism to extract local and non-local information from images via different transformer encoders and CNNs, establishing a stronger connection between subjective and objective assessments through sorting within batches of images based on relative distance information, and a self-consistency approach to self-supervision to address the degradation of no-reference image quality assessment models under equivariant transformations.
PromptIQA can adapt to new requirements without fine-tuning, outperforming existing methods.
PromptIQA introduces a novel approach to adapt to new assessment requirements without fine-tuning, using prompts for targeted predictions. The model outperforms existing methods by learning diverse requirements and achieving better generalization.