Abstract:Recent research on deep convolutional neural networks (CNNs) has provided a significant performance boost on efficient super-resolution (SR) tasks by trading off the performance and applicability. However, most existing methods focus on subtracting feature processing consumption to reduce the parameters and calculations without refining the immediate features, which leads to inadequate information in the restoration. In this paper, we propose a lightweight network termed DDistill-SR, which significantly improves the SR quality by capturing and reusing more helpful information in a static-dynamic feature distillation manner. Specifically, we propose a plug-in reparameterized dynamic unit (RDU) to promote the performance and inference cost trade-off. During the training phase, the RDU learns to linearly combine multiple reparameterizable blocks by analyzing varied input statistics to enhance layer-level representation. In the inference phase, the RDU is equally converted to simple dynamic convolutions that explicitly capture robust dynamic and static feature maps. Then, the information distillation block is constructed by several RDUs to enforce hierarchical refinement and selective fusion of spatial context information. Furthermore, we propose a dynamic distillation fusion (DDF) module to enable dynamic signals aggregation and communication between hierarchical modules to further improve performance. Empirical results show that our DDistill-SR outperforms the baselines and achieves state-of-the-art results on most super-resolution domains with much fewer parameters and less computational overhead. We have released the code of DDistill-SR at <a class="link-external link-https" href="https://github.com/icandle/DDistill-SR" rel="external noopener nofollow">this https URL</a>.

Classification-Based Dynamic Network for Efficient Super-Resolution

Orientation-Aware Deep Neural Network for Real Image Super-Resolution.

Image super-resolution via dynamic network

Dynamic image super-resolution via progressive contrastive self-distillation

Deep Differential Convolutional Network for Single Image Super-Resolution.

DDistill-SR: Reparameterized Dynamic Distillation Network for Lightweight Image Super-Resolution

Lightweight Progressive Residual Clique Network for Image Super-Resolution.

A Dual CNN for Image Super-Resolution

Accelerating Image Super-Resolution Networks with Pixel-Level Classification

Learning Frequency-aware Dynamic Network for Efficient Super-Resolution.

Accurate image super-resolution using dense connections and dimension reduction network

Image Super-Resolution Using Deep Convolutional Networks

Dynamic Dual Attention Iterative Network for Image Super-Resolution

A Compact Deep Neural Network for Single Image Super-Resolution

Super-resolution Guided Knowledge Distillation for Low-Resolution Image Classification

A Novel Multi-Scale Adaptive Convolutional Network for Single Image Super-Resolution.

Image Super-Resolution Based on Dual Path Network

Sub-Pixel Convolutional Neural Network for Image Super-Resolution Reconstruction

Dual-attention Guided Multi-Scale Network for Single Image Super-Resolution

Image Super-Resolution Via a Novel Cascaded Convolutional Neural Network Framework.

Adaptive Dynamic Shuffle Convolutional Parallel Network for Image Super-Resolution