MobileSR

---

Introduction

We describe an efficient super-resolution network, MobilSR, and proposes a newly-devised convolution, parallel- group convolution. Parallel-group convplution divides standard convolutions into four groups and introduces a shortcut of 1×1 convolution. It achieves approximate performance but reduces the number of parameters by four times compared to a standard convolution.

• parallel group convolution
• performance

Code

Clone this repository into any place you want.

git clone https://github.com/DestinyK/MobileSR
cd MobileSR

We refer to this EDSR code

Datasets

We used DIV2K dataset to train our model. Please download it from here (7.1GB).

You can evaluate your models with widely-used benchmark datasets:

Set5 - Bevilacqua et al. BMVC 2012,

Set14 - Zeyde et al. LNCS 2010,

B100 - Martin et al. ICCV 2001,

Urban100 - Huang et al. CVPR 2015.

Start

cd src
sh demo.sh

You can find the result images from experiment/test/results folder.