Image-Difficulty-Aware Evaluation of Super-Resolution Models
Authors: Atakan Topaloglu, Ahmet Bilican, Cansu Korkmaz, A. Murat Tekalp
Venue: IEEE International Conference on Image Processing (ICIP) Workshops, 2025
Overview
This work proposes an image-difficulty-aware evaluation framework for super-resolution models, providing more nuanced assessment of model performance across different image complexities. The framework accounts for the inherent difficulty of different images, leading to more meaningful and fair comparisons between models.
Key Contributions
- Image-difficulty-aware evaluation framework
- Novel metrics for assessing super-resolution performance
- More fair and meaningful model comparisons