KBSMosaic

Educational project to solve mosaic instances with NeurASP

Generator

python generator/mosaicGenerator <file path> <number of instances>

For example:

python generator/mosaicGenerator ../instances/ 10 (To create images and instance files)
python generator/mosaicGenerator instances/data.csv 10000 (To create csv file)

Train NeurASP

To train a new model:

python trainNASP/train.py <optional number of data>

Trained model will be saved in root directory as model_dataXX.pt.

Test a given model:

python trainNASP/test.py

Use inference to get a solution:

python trainNASP/infer.py <IMAGE>

Train a pure CNN

Needs a csv Dataset and will create a model.h5 file in the root directory. If a model is given, it wont train a new one.

python trainNN/mosaic.py <dataset> <optional model>