This package contains several functions for loading data, running experiments and visualizing results.
| Module | Contents |
|---|---|
| dataloading | functions for loading single patches (load_patches) or the whole FAUST and GALLOP training and testing datasets (load_train_test_data_faust, load_train_test_data_gallop) and for converting a dataset into a PyTorch Geometric dataset (get_pygeo_dataset) |
| experiment_runner | a class (ExperimentRunner) for running and saving experiments for multiple different models for multiple random seeds |
| metrics | a PyTorch implementation (r2_score) of the |
| model_trainer | a class (ModelTrainer) for training and testing a given model on a given dataset |
| plotting | a function (get_mesh_predictions) to compute the nodewise mesh predictions of a given model as well as multiple functions for nicely visualising meshes or patches |
| normalize | a function (normalize) to normalize input meshes |
| utils | functions to create directories and truncate colormaps |
| srmesh_dataset_specnet | a class (srmesh_dataset_specnet) to evaluate the results |