Softwarepraktikum (25SS)
Thema | Betreuer(in) |
---|---|
Next generation Video Platform
Learn how Netflix, YouTube & Co works and contribute to the next generation Video
Platform in the context of the ATHENA project.
|
C. Timmerer |
Python API for C++ library in the scope of mobile robots
We are looking for a dedicated bachelor student to develop a Python API for our prize-winning C++ library for state estimation in robotics (c.f. github repository for more details). The library is used to fuse real-time sensor data on mobile robots for robust localization in a recursive manner. It achieves true modularity by de-coupling the sensor states from the essential navigation states allowing for adding or removal of sensors at runtime. The Python API shall act as a wrapper to the underlying C++ implementation exposing the main top level functions. This will facilitate the adoption of our library to novel sensor and robotic platforms and speed up the integration by being able to work from simple Python scripts or even the command line. The main tasks of the project are • Generation of a Python API for the main functions of the C++ library • Comprehensive tests of the Python API for different use cases |
J. Steinbrenner P. Schartner |
Optimizing Access Networks for Low-Latency Real-Time Applications
Choosing the right access network is critical to ensuring low-latency performance for real-time applications, particularly in the context of the computing continuum. The decision impacts application responsiveness and the seamless integration of edge, fog, and cloud computing resources. This study explores the implications of access network selection, highlighting strategies to optimize latency and meet real-time requirements.
|
K. Horvath |
LLM-Agents im Smart Home
Im Gegensatz zu LLM-basierten Chatbots wie ChatGPT können LLM-Agents selbstständig Informationen aus der Umgebung abrufen und eigenständig Aktionen ausführen. Dieses Projekt untersucht, inwieweit LLM-Agents die Konfiguration und Nutzung heterogener Smart-Home-Systeme für Laien vereinfachen können.
|
G. Leitner J. Köpke |
Der Anti-Metamorphopsie-Visor von/für Geordi La Forge 2.0 | P. Schartner |
Implementing End-to-end encrypted HLS video streaming
Current encryption solutions for Video streaming focus solely on DRM (digital rights management) protection goals. Zeitkapsl (a young start-up working on protecting photos/videos) is currently working on an end-to-end encrypted video streaming schema for various platforms (Android, iOS and Web) that enables full privacy over the video content and metadata. Keywords: Android, iOS, HLS, Service Workers, Web Assembly, AES-256-GCM, ffmpeg |
P. Spiess-Knafl P. Schartner |
Privacy friendly face/object detection on Web/Android/iOS
In order to enable customers with a comfortable and privacy-preserving photo/video search by person names and faces, there is a need to perform facial and object detection in photos/videos on constrained mobile devices. Sending any (unencrypted) photos/videos to a cloud infrastructure is no option. Zeitkapsl (a young start-up working on protecting photos/videos) needs support in implementing innovative and performant ideas on how to work in constraint mobile and browser based environment using quantization of machine-learning models in order to extract facial and object information from photos and videos. Keywords: Android, iOS, WASM, object detection, face detection, image retrieval, video retrieval, CoreML, Open CV, ONNX Runtime, AES-256-GCM |
P. Spiess-Knafl P. Schartner |
KI-unterstützte Bearbeitung von Begünstigtenanfragen
Viele Begünstigte von Krankheits- und Ausbildungsleistungen benötigen professionelle Unterstützung bei der Geltendmachung ihrer Ansprüche. Spezialisierte Dienstleister bieten daher Beratung zum Vorgehen und Hilfe bei der Leistungsbeantragung an. Dieser Tätigkeitsbereich geht mit einem hohem Aufkommen an E-Mail Anfragen einher, für die jeweils geeignete Auskünfte und Informationen zum weiteren Vorgehen bereitgestellt werden müssen. Das Ziel dieses Softwarepraktikums besteht darin, in Zusammenarbeit mit einem Kärntner Dienstleister zu untersuchen, inwieweit KI-Unterstützung die Analyse, Kategorisierung und den Entwurf von Antwortschreiben an Begünstigte effektiver gestalten kann.
|
M. Gebser |
More to come… | |
Eigene Themen sind willkommen – Rahmenbedingungen siehe oben! |