Uncertainty Estimation for Super-Resolution using ESRGAN
This work addresses the problem of unreliable super-resolution outputs for users by providing uncertainty estimates, though it is incremental as it applies existing uncertainty methods to a specific domain.
The paper tackles the lack of principled uncertainty estimation in deep learning-based image super-resolution models like ESRGAN by enhancing them with Monte Carlo-Dropout and Deep Ensemble methods, resulting in decently calibrated uncertainty estimates without performance drop.
Deep Learning-based image super-resolution (SR) has been gaining traction with the aid of Generative Adversarial Networks. Models like SRGAN and ESRGAN are constantly ranked between the best image SR tools. However, they lack principled ways for estimating predictive uncertainty. In the present work, we enhance these models using Monte Carlo-Dropout and Deep Ensemble, allowing the computation of predictive uncertainty. When coupled with a prediction, uncertainty estimates can provide more information to the model users, highlighting pixels where the SR output might be uncertain, hence potentially inaccurate, if these estimates were to be reliable. Our findings suggest that these uncertainty estimates are decently calibrated and can hence fulfill this goal, while providing no performance drop with respect to the corresponding models without uncertainty estimation.