cv
Basics
Name | Anna Penzkofer |
Label | PhD Candidate |
anna.penzkofer@vis.uni-stuttgart.de | |
Url | https://apenzko.github.io/ |
Work
- 2023.02 - present
Stuttgart, Germany
PhD Candidate
Institute for Visualisation and Interactive Systems, University of Stuttgart
Part of research group Human-Computer-Interaction and Cognitive Systems led by Prof. Dr. Andreas Bulling
- Interactive Behaviour
- Computer Vision
- Reinforcement Learning
- 2021.02 - 2022.03
Munich, Germany
Working Student -- Perception & Cognition
German Aerospace Center (DLR)
Part of perception & cognition group, working on semantic segmentation for real-time navigation robot Rollin' Justin. Collaboration on open-source tool Blender-Proc https://github.com/DLR-RM/BlenderProc.
- 2019.10 - 2020.12
Munich, Germany
Working Student -- Data Analytics
Siemens AG
Preprocess and analyze project data, implement data augmentation and fit forecasting for classification models.
- 2018.11 - 2019.08
Stuttgart, Germany
Research Assistant -- Machine Learning
University of Stuttgart
Conduct research on learning-based gaze estimation as well as image synthesis such as gaze redirection and generative adversarial networks, including testing state-of-the-art methods, analyzing and developing new models.
Education
-
2020.04 - 2022.12 Munich, Germany
Masters
Technical University Munich
Computer Science
- Cognitive Systems
- Machine Learning for Computer Vision
- Reinforcement Learning for Robotics
-
2017.10 - 2020.03 Stuttgart, Germany
Bachelors
University of Stuttgart
Computer Science
- Artificial Intelligence
- Machine Learning
- Computer Vision
Certificates
Reinforcement Learning | ||
University of Alberta | 2024-02-26 |
Publications
-
Oct 2021 ConAn: A Usable Tool for Multimodal Conversation Analysis
International Conference on Multimodal Interaction (ICMI)
ConAn is a software tool that integrates multimodal analyses on 360 degree videos of conversations and displays them in a useable graphical user interface (GUI). Modalities include eye gaze, body pose, facial expressions, speaking activity and object tracking.
-
June 2023 Int-HRL: Towards Intention-based Hierarchical Reinforcement Learning
Adaptive and Learning Agent Workshop (ALA)
Proposes new method Intention-based Hierarchical Reinforcement Learning (Int-HRL). Tested on the long-horizon, sparse reward task Montezuma's Revenge from the Atari2600 game suite.
Languages
German | |
Native speaker |
English | |
Fluent |