Awesome

Compressed Sensing using Generative Models

This repository provides code to reproduce results from the paper: Compressed Sensing using Generative Models.

Here are a few example results:

Reconstruction	Super-resolution	Inpainting
<img src="https://github.com/AshishBora/csgm/blob/master/example_results/celebA_reconstr_500_orig_lasso-dct_lasso-wavelet_dcgan.png" alt="celebA_reconstr" width="300">	<img src="https://github.com/AshishBora/csgm/blob/master/example_results/celebA_superres_orig_blurred_dcgan.png" alt="celebA_superres" width="300">	<img src="https://github.com/AshishBora/csgm/blob/master/example_results/celebA_inpaint_orig_blocked_dcgan.png" alt="celebA_inpaint" width="300">
<img src="https://github.com/AshishBora/csgm/blob/master/example_results/mnist_reconstr_100_orig_lasso_vae.png" alt="mnist_reconstr" width="300">	<img src="https://github.com/AshishBora/csgm/blob/master/example_results/mnist_superres_orig_blurred_vae.png" alt="mnist_superres" width="300">	<img src="https://github.com/AshishBora/csgm/blob/master/example_results/mnist_inpaint_orig_blocked_vae.png" alt="mnist_inpaint" width="300">

Here we show the evolution of the reconstructed image for different number of iterations:

Steps to reproduce the results

NOTE: Please run all commands from the root directory of the repository, i.e from csgm/

Requirements:

Python 2.7
Tensorflow 1.0.1
Scipy
PyPNG
(Optional : for lasso-wavelet) PyWavelets
(Optional) CVXOPT

Pip installation can be done by $ pip install -r requirements.txt

Preliminaries

Clone the repository and dependencies

$ git clone https://github.com/AshishBora/csgm.git
$ cd csgm
$ git submodule update --init --recursive

Download/extract the datasets:
```
$ ./setup/download_data.sh
```
Download/extract pretrained models or train your own!
- To download pretrained models: $ ./setup/download_models.sh
- To train your own
  - VAE on MNIST: $ ./setup/train_mnist_vae.sh
  - DCGAN on celebA, see https://github.com/carpedm20/DCGAN-tensorflow
To use wavelet based estimators, you need to create the basis matrix:
```
$ python ./src/wavelet_basis.py
```

Demos

The following are the supported experiments and example commands to run the demos:

[Note: For celebA experiments, we run very few iterations per experiment (30) to give a quick demo. To get results with better quality, increase the number of iterations to at least 500 and use at least 2 random restarts.]

Reconstruction from Gaussian measurements
- $ ./quick_scripts/mnist_reconstr.sh
- $ ./quick_scripts/celebA_reconstr.sh "./images/182659.jpg"
Super-resolution
- $ ./quick_scripts/mnist_superres.sh
- $ ./quick_scripts/celebA_superres.sh "./images/182659.jpg"
Reconstruction for images in the span of the generator
- $ ./quick_scripts/mnist_genspan.sh
- $ ./quick_scripts/celebA_genspan.sh
Quantifying representation error
- $ ./quick_scripts/mnist_projection.sh
- $ ./quick_scripts/celebA_projection.sh "./images/182659.jpg"
Inpainting
- $ ./quick_scripts/mnist_inpaint.sh
- $ ./quick_scripts/celebA_inpaint.sh "./images/182659.jpg"

Reproducing quantitative results

Create a scripts directory $ mkdir scripts
Identfy a dataset you would like to get the quantitative results on. Locate the file ./quant_scripts/{dataset}_reconstr.sh.
Change BASE_SCRIPT in src/create_scripts.py to be the same as given at the top of ./quant_scripts/{dataset}_reconstr.sh.
Optionally, comment out the parts of ./quant_scripts/{dataset}_reconstr.sh that you don't want to run.
Run ./quant_scripts/{dataset}_reconstr.sh. This will create a bunch of .sh files in the ./scripts/ directory, each one of them for a different parameter setting.
Start running these scripts.
- You can run $ ./utils/run_sequentially.sh to run them one by one.
- Alternatively use $ ./utils/run_all_by_number.sh to create screens and start proccessing them in parallel. [REQUIRES: gnu screen][WARNING: This may overwhelm the computer]. You can use $ ./utils/stop_all_by_number.sh to stop the running processes, and clear up the screens started this way.
Create a results directory : $ mkdir results. To get the plots, see src/metrics.ipynb. To get matrix of images (as in the paper), run $ python src/view_estimated_{dataset}.py.
You can also manually access the results saved by the scripts. These can be found in appropriately named directories in estimated/. Directory name conventions are defined in get_checkpoint_dir() in src/utils.py

Miscellaneous

For a complete list of images not used while training on celebA, see here.