CSNet-Pytorch

Pytorch code for paper

• "Deep Networks for Compressed Image Sensing" ICME2017

• "Image Compressed Sensing Using Convolutional Neural Network" TIP2019

Requirements and Dependencies

• Ubuntu 16.04 CUDA 10.0
• Python3 （Testing in Python3.5）
• Pytorch 1.1.0
• Torchvision 0.2.2

Details of Implementations

In our code, two model version are included:

• simple version of CSNet (Similar with paper ICME2017)
• Enhanced version of CSNet (local skip connection + global skip connection + resudial learning)

How to Run

Training CSNet

• Preparing the dataset for training

• Editing the path of training data in file train.py.

• For CSNet training in terms of subrate=0.1:

python train.py --sub_rate=0.1 --block_size=32

Testing CSNet

• Preparing the dataset for testing

• Editing the path of trained model in file test.py and test_new.py.

• For CSNet testing in terms of subrate=0.1: (ps: For this testing code, there is a big gap compared with the result in the publised paper. And I am confused about it. If you know the reason, please let me know. Thanks very much!)

python test.py --sub_rate=0.1 --block_size=32

• For CSNet testing (new testing code) in terms of subrate=0.1:

python test_new.py --cuda --sub_rate=0.1 --block_size=32

CSNet results

Subjective results

Objective results

Additional instructions

• For training data, you can choose any natural image dataset.
• The training data is very important, if you can not achieve ideal result, maybe you can focus on the augmentation of training data or the structure of the network.
• If you like this repo, Star or Fork to support my work. Thank you.
• If you have any problem for this code, please email: wenxuecui@stu.hit.edu.cn