This notebook covers the Python implementation of a generative adversarial network (GAN) for enhancing the resolution of particle-image-velocimetry (PIV) images from incomplete high-resolution pairs. Particle tracking velocimetry is the experimental technique used to acquire those incomplete high-resolution images. The article associate to his work can be found at [to be published].
Four different cases are available:
- Cylinder wake: direct-numerical simulation (DNS) data generated from Taira and Colonius (2007) and Kutz et al. (2016).
- Turbulent-channel flow: DNS data from a turbulent channel flow with friction Reynolds number
$Re_{\tau}=1000$ available at Johns Hopkins Turbulence Database. - Turbulent boundary layer flow: experimental data of a turbulent boundary layer with friction Reynolds number
$Re_{\tau}\approx 900$ acquired in the water-tunnel facility at Universidad Carlos III de Madrid. - Blunt-body wake: experimental data of the flow around a blunt body acquired in the wind-tunnel facility at Universidad Carlos III de Madrid.
Use the package manager pip to install the required dependencies.
pip install -r requirements.txtTo generate the tfrecord files, execute:
guest@vegeta:~$ python run_generate_tfrecords.py -c channel -u 4To run the training procedure, execute:
guest@vegeta:~$ python run_training --case channel --upsampling 4 --model_name architecture01 --learning_rate 1e-4To compute the prediction of the testing dataset, execute:
guest@vegeta:~$ python run_predictions -c channel -u 4 -m architecture01 -l 1e-4This repository has been used for the following scientific publications:
To be a nnounced
This repository has been developed in the Experimental Aerodynamics and Propulsion group at Universidad Carloss III de Madrid. The following researches and students are acknowledged for their contributions:
- Alejandro Güemes
- Stefano Discetti
- Carlos Sanmiguel
Pull requests are welcome. For major changes, please open an issue first to discuss what you would like to change.
Please make sure to update tests as appropriate.