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.