If you want to have more information about our implementaiton please look at project description in the pdf file.
To get the training data please execute the mask_creator.py
script but first create an empty folder called dataset
and put the templates under images/templates
and the backgrounds under images/templates
.
To train the model execute the file model.py
.
To test the model execute the eval_model.py
file. This script will take the first element inside dataset/images
and it will output the annotated image in the root directory.
Each template is annotated with a different color.
If you have any doubts please contact us at noli.manzoni@usi.ch or michael.denzler@usi.ch