-
FilmBibWeb
This was the first larger project I coded by myself at the age of 17 in order to help my dad structure and organize the loads of recorded movies and series he had collected over the years. This project is based on a Raspberry Pi running a local Apache web server and managing the MariaDB that I created to keep track of the movies and their storage locations. The web interface which is based on PHP allows my dad to quickly check for already recorded movies and their current storage location. The scanning of the drives is done by a python script, which also adds additional information to each movie entry by calling the IMDb-API.
- PHP
- Python
- MariaDB
- Raspberry Pi
- HTML, CSS, JS
-
Spotify-Playlist-Creator
I created this app/website because of my love for discovering new music and the process of finding new songs I can play and map for Beat Saber. The website is based on the Svelte-Framework and designed with Tailwind. The basic functionality is to create playlists you can add to Spotify that are filled with random songs based on certain features like popularity or genre. The song search is based on the Spotify API and bound to your own account and preferences. Therefore, the results tend to feel less random than I would have liked to. Since the app has not been verified by Spotify, the functionality can only be accessed by me and persons I add to the developer list.
- SpotifyAPI
- HTML, CSS, TypeScript
- Svelte
- Tailwind
-
VR-Avatar-Reconstruction
This is the project I have done for my bachelor thesis "Reconstruction of virtual reality avatars from sparse data using deep learning." The goal was to recreate full body avatars from the sparse inputs of a VR headset, and comparing multiple deep learning techniques in doing so. Therefore, I trained several Transformer-, LSTM- and CNN-Architectures on the "Talking With Hands 16.2M"-dataset and compared their results in predicting the next full body pose from only the positions and rotations of headset and two controllers. In the end no architecture was able to consistently predict the lower body movements from the upper body input signals, but the Transformer-Architecture showed the most stable and precise results.
- Python
- Keras
- Numpy
- Matplotlib
- PyMO
- Google Colab
-
Personal Website
This is the website you are currently on. It is only built on HTML and Bootstrap and should therefore be very fast and responsive. I hope this website could give you some insights about myself and what I am doing :D. Feel free to contact me on the socials that were linked on the home page.
- HTML
- Bootstrap
- Responsive Webdesign
