Event-guided Multi-patch Network with Self-supervision for Non-uniform Motion Deblurring

TL;DR

This paper introduces a novel event-guided multi-patch network with self-supervision that significantly improves non-uniform motion deblurring, achieving real-time performance on high-resolution videos and robustness to transformations and noise.

Contribution

It proposes a hierarchical multi-patch network with event guidance and self-supervision, advancing deblurring performance and robustness over prior models.

Findings

Achieves state-of-the-art results on GoPro and VideoDeblur datasets.

Runs in real-time at 30ms per 720p image at 30fps.

Improves deblurring quality by over 1.2dB with increased network depth.

Abstract

Contemporary deep learning multi-scale deblurring models suffer from many issues: 1) They perform poorly on non-uniformly blurred images/videos; 2) Simply increasing the model depth with finer-scale levels cannot improve deblurring; 3) Individual RGB frames contain a limited motion information for deblurring; 4) Previous models have a limited robustness to spatial transformations and noise. Below, we extend the DMPHN model by several mechanisms to address the above issues: I) We present a novel self-supervised event-guided deep hierarchical Multi-patch Network (MPN) to deal with blurry images and videos via fine-to-coarse hierarchical localized representations; II) We propose a novel stacked pipeline, StackMPN, to improve the deblurring performance under the increased network depth; III) We propose an event-guided architecture to exploit motion cues contained in videos to tackle complex blur in videos; IV) We propose a novel self-supervised step to expose the model to random transformations (rotations, scale changes), and make it robust to Gaussian noises. Our MPN achieves the state of the art on the GoPro and VideoDeblur datasets with a 40x faster runtime compared to current multi-scale methods. With 30ms to process an image at 1280x720 resolution, it is the first real-time deep motion deblurring model for 720p images at 30fps. For StackMPN, we obtain significant improvements over 1.2dB on the GoPro dataset by increasing the network depth. Utilizing the event information and self-supervision further boost results to 33.83dB.