Take a Prior from Other Tasks for Severe Blur Removal

TL;DR

This paper introduces a novel method for severe blur removal that leverages priors learned from other high-level vision tasks through a cross-level feature learning strategy and semantic prior embedding, improving deblurring performance.

Contribution

It proposes a cross-level feature learning strategy based on knowledge distillation and a semantic prior embedding layer to enhance severe blur removal in various models.

Findings

Improved deblurring results on GoPro and RealBlur datasets.

The method generalizes well across different models and real-world images.

Significant enhancement over baseline deblurring methods.

Abstract

Recovering clear structures from severely blurry inputs is a challenging problem due to the large movements between the camera and the scene. Although some works apply segmentation maps on human face images for deblurring, they cannot handle natural scenes because objects and degradation are more complex, and inaccurate segmentation maps lead to a loss of details. For general scene deblurring, the feature space of the blurry image and corresponding sharp image under the high-level vision task is closer, which inspires us to rely on other tasks (e.g. classification) to learn a comprehensive prior in severe blur removal cases. We propose a cross-level feature learning strategy based on knowledge distillation to learn the priors, which include global contexts and sharp local structures for recovering potential details. In addition, we propose a semantic prior embedding layer with multi-level aggregation and semantic attention transformation to integrate the priors effectively. We introduce the proposed priors to various models, including the UNet and other mainstream deblurring baselines, leading to better performance on severe blur removal. Extensive experiments on natural image deblurring benchmarks and real-world images, such as GoPro and RealBlur datasets, demonstrate our method's effectiveness and generalization ability.