Reading: MSRN — Multi-Scale Residual Network (Super Resolution)

On Par With EDSR But Much Fewer Number of Parameters, Outperforms LapSRN, DRCN, VDSR, ESPCN, FSRCNN, SRCNN

Outline

1. MSRN: Network Architecture

MSRN: Network Architecture

2. Multi-Scale Residual Block (MSRB)

Multi-Scale Residual Block (MSRB)

2.1. Multi-Scale Features Fusion

2.2. Local Residual Learning

3. Hierarchical Feature Fusion Structure (HFFS)

4. Image Reconstruction

Different Image Reconstruction Modules
Detailed configuration information about the reconstruction structure.

5. SOTA Comparison

Quantitative comparisons of state-of-the-art methods. Red text indicates the best performancen and blue text indicate the second best performance.
Visual comparison

6. Further Study

6.1. Benefit of MSRB

Quantitative comparison of three different feature extraction blocks
Feature maps visualization.

6.2. Benefit of Increasing The Number of MSRB

Performance comparison of MSRN with different number of MSRBs.

6.3. Other Tasks

Application examples for image denoising and image dehazing, respectively.

PhD, Researcher. I share what I've learnt and done. :) My LinkedIn: https://www.linkedin.com/in/sh-tsang/, My Paper Reading List: https://bit.ly/33TDhxG