This is the implementation of the SuperMeshingNet, for paper: SuperMehsingNet: A Novel Method for Boosting the Mesh Density in Numerical Computation within 2D Domain
The architecture of the SuperMeshingNet is shown in figure.
To train the model, please download the project and run:
python code/train.py
To evaluate the model, please run:
python code/test.py
If you want to test different scaling factors, please change the --scale argument in /code/train.py and /code/test.py, replacing 4 as 2 or 8.
The result floder includes the reslt of experiment.
