On-Demand Learning for Deep Image Restoration
This addresses a domain-specific problem for image restoration practitioners by improving generalization, though it is incremental as it builds on existing deep learning methods.
The paper tackles the problem of training deep image restoration models that generalize across varying levels of corruption, such as noise or blur, by proposing an on-demand learning algorithm that uses feedback to self-generate training instances. The result shows consistent outperformance over baseline methods on four restoration tasks and three datasets.
While machine learning approaches to image restoration offer great promise, current methods risk training models fixated on performing well only for image corruption of a particular level of difficulty---such as a certain level of noise or blur. First, we examine the weakness of conventional "fixated" models and demonstrate that training general models to handle arbitrary levels of corruption is indeed non-trivial. Then, we propose an on-demand learning algorithm for training image restoration models with deep convolutional neural networks. The main idea is to exploit a feedback mechanism to self-generate training instances where they are needed most, thereby learning models that can generalize across difficulty levels. On four restoration tasks---image inpainting, pixel interpolation, image deblurring, and image denoising---and three diverse datasets, our approach consistently outperforms both the status quo training procedure and curriculum learning alternatives.