Image Super-Resolution use caffe.
This project implements two articles，"Image Super-Resolution Using Deep Convolutional Networks"(ECCV 2014) and "Accurate Image Super-Resolution Using Very Deep Convolutional Networks" (CVPR 2016 Oral Paper).
├── dataset (数据集)
│ ├── hdf5
│ │ ├── SRCNN-5
│ │ ├── SRCNN-Pool
│ │ └── VDSR-20
│ └── images
│ ├── 291
│ ├── 91
│ ├── 91-aug
│ └── test
├── external
│ └── caffe
├── pdf
├── test （测试评估）
│ ├── images
│ │ ├── Set14
│ │ └── Set5
│ ├── matconvnet
│ ├── results
│ │ ├── SRCNN-5
│ │ ├── SRCNN-Pool
│ │ └── VDSR-20
│ │ ├── official
│ │ └── ours
│ └── utils
└── train （训练）
├── SRCNN-5
│ ├── caffemodel
│ ├── convert
│ ├── demo
│ │ └── utils
│ └── logs
├── SRCNN-Pool
│ ├── caffemodel
│ ├── convert
│ ├── logs
│ └── prototxt
│ └── data_pool
│ └── logs
└── VDSR-20
├── caffemodel
├── convert
└── logs下载数据集 zip(train) zip(test)
git clone https://github.com/mindcont/SR-Caffe.git
cd SR-Caffe && mkdir dataset
wget http://cv.snu.ac.kr/research/VDSR/train_data.zip
wget http://cv.snu.ac.kr/research/VDSR/test_data.zip
unzip train_data.zip images/
# use matlab
run train/data_aug.mgenerate_train.m
generate_test.mrun train_VDSR-20.sh or train_SRCNN-5.shrun demo_SR.m-
"Image super-resolution as sparse representation of raw image patches"(CVPR2008)
-
Single Image Super-Resolution from Transformed Self-Exemplars(CVPR2015)
-
"Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Network"(arxiv, 21 Nov, 2016)
-
"Deep Laplacian Pyramid Networks for Fast and Accurate Super-Resolution"(CVPR 2017)