This repository contains the document and code for my master thesis at Saarland University.
Structure of repository
--plots
--src
-- dataset
-- models
-- training files
Dataset folder consists of dataset classes for single-view and multi-view models.
Models folder consist of both the model class and architecture files.
To Train a model and save checkpoints, one can select from the train_modality.py files
Eval_modality.py evaluates trained model on independent test set and outputs metrics, plots and metadata files.