FSRCNN

This repository is implementation of the "Accelerating the Super-Resolution Convolutional Neural Network".

Differences from the original

• Added the zero-padding
• Used the Adam instead of the SGD

Requirements

• PyTorch 1.0.0
• Numpy 1.15.4
• Pillow 5.4.1
• h5py 2.8.0
• tqdm 4.30.0

Train

The 91-image, Set5 dataset converted to HDF5 can be downloaded from the links below.

Dataset	Scale	Type	Link
91-image	2	Train	Download
91-image	3	Train	Download
91-image	4	Train	Download
Set5	2	Eval	Download
Set5	3	Eval	Download
Set5	4	Eval	Download

Otherwise, you can use prepare.py to create custom dataset.

python train.py --train-file "BLAH_BLAH/91-image_x3.h5" \
                --eval-file "BLAH_BLAH/Set5_x3.h5" \
                --outputs-dir "BLAH_BLAH/outputs" \
                --scale 3 \
                --lr 1e-3 \
                --batch-size 16 \
                --num-epochs 20 \
                --num-workers 8 \
                --seed 123                

Test

Pre-trained weights can be downloaded from the links below.

Model	Scale	Link
FSRCNN(56,12,4)	2	Download
FSRCNN(56,12,4)	3	Download
FSRCNN(56,12,4)	4	Download

The results are stored in the same path as the query image.

python test.py --weights-file "BLAH_BLAH/fsrcnn_x3.pth" \
               --image-file "data/butterfly_GT.bmp" \
               --scale 3

Results

PSNR was calculated on the Y channel.

Set5

Eval. Mat	Scale	Paper	Ours (91-image)
PSNR	2	36.94	37.12
PSNR	3	33.06	33.22
PSNR	4	30.55	30.50

Original	BICUBIC x3	FSRCNN x3 (34.66 dB)
Original	BICUBIC x3	FSRCNN x3 (28.55 dB)