This repository is associated with my Master Thesis at the Artificial Intelligence and Machine Learning (AIML) Group in the Department of Informatics at the University of Zurich. In this thesis, I conduct a systematic comparison of four relevant Open Set algorithms. The repository mirrors the original project structure on the IfI server rolf and consists of the following directories:
-
code: contains the source code that builds the foundation of the comparison of the algorithms. This source code is organised in form of two Python packages (eosa and imageNet_protocols). For more information on those packages as well as on running the experiments, please refer to the directory code.
-
data: most importantly, this directory hosts the protocol-specific csv files that contain the access information necessary for loading the datasets. These files are not uploaded to GitLab as they can simply be generated by running the script prepare_protocols.py.
-
results: comprises various files that are linked to the final results. For more information on these files, please refer to the respective subdirectories.