Code base for documenting experiments conducted during a 5th semester project at Aalborg University (AAU). The project is experimenting on how to generate images of cars having certain features. The project concerns the theme: Synthetic Data Generation, as it tries to generate synthetic product images that should be used to better display products that do not have enough images for showcasing.
Each folder is related to an experiment conducted in the project. This is with the exception of a few folders which contain the used images for training the models.
To run an experiment change your current directory to a folder containing a main.py-file, which should be run using python. For common systems the command should look like:
python main.py or
python3 main.py.
The experiments has been logged using Weight and Biases, which can be found as their website wandb.ai. To log graphs and other data according to the presented results in the project report it is necessary to create a wandb account and change any line in the main.py-files concerning wandb logging.
Example:
wandb.init(project='<project name>', entity='<Entity name>', name='<Name of run>', notes='<Notes>')
If the line has been changed to use your own wandb account, the command for running the experiments will be:
python main.py -w 1 or
python3 main.py -w 1.
The project uses python 3.9.0.
The dependencies with versions can be found in requirements.txt.