Full Resolution Image Compression with Recurrent Neural Networks ArXiv:1608.05148

Check my repo here for the original Tensorflow version deployed and created by Google Researcher Toderici with a Colab Version

Requirements

• PyTorch 0.2.0

Train

python train.py -f /path/to/your/images/folder/like/mscoco

Encode and Decode

Encode

python encoder.py --model checkpoint/encoder_epoch_00000066.pth --input /path/to/your/example.png --cuda --output /path/to/output/folder/example.npz --iterations 16

This will output binary codes saved in .npz format.

Decode

python decoder.py --model checkpoint/decoder_epoch_00000066.pth --input /path/to/your/example.npz --cuda --output /path/to/output/folder

This will output images of different quality levels.

Test

Get Kodak dataset

bash test/get_kodak.sh

Encode and decode with RNN model

bash test/enc_dec.sh

Encode and decode with JPEG (use convert from ImageMagick)

bash test/jpeg.sh

Calculate SSIM

bash test/calc_ssim.sh

Draw rate-distortion curve

python test/draw_rd.py

Result

LSTM (Additive Reconstruction), before entropy coding

Rate-distortion

kodim10.png

Original Image

Below Left: LSTM, SSIM=0.865, bpp=0.125

Below Right: JPEG, SSIM=0.827, bpp=0.133

Below Left: LSTM, SSIM=0.937, bpp=0.250

Below Right: JPEG, SSIM=0.918, bpp=0.249

Below Left: LSTM, SSIM=0.963, bpp=0.375

Below Right: JPEG, SSIM=0.951, bpp=0.381

What's inside

• train.py: Main program for training.
• encoder.py and decoder.py: Encoder and decoder.
• dataset.py: Utils for reading images.
• metric.py: Functions for Calculatnig MS-SSIM and PSNR.
• network.py: Modules of encoder and decoder.
• modules/conv_rnn.py: ConvLSTM module.
• functions/sign.py: Forward and backward for binary quantization.

Official Repo

https://github.com/tensorflow/models/tree/master/compression