On January 15, 2025, Jasmin Gründling-Riener, TU Wien Vice-Rector for Academic Affairs, and Daniel Löcker from the city’s Department of Cultural Affairs, presented the Diploma Thesis Awards 2024 of the City of Vienna at TU Wien. Christian Stippel won the City of Vienna Diploma Thesis Award, endowed with 750 Euros, for his thesis “Multimodal machine learning to alleviate data scarcity: trimodal dataset generation”. He was supervised by Martin Kampel and co-supervised by Thomas Heitzinger.

In his thesis, Christian Stippel investigates how datasets for human behavior analysis in computer vision, which mostly rely on Red-Green-Blue (RGB) imaging, can be augmented with thermal and depth data to make them more robust. To address challenges with RGB imaging, like lighting conditions and privacy concerns, he presents TRISTAR. TRISTAR is a public archive that includes synchronized RGB, depth, and thermal data collected in various environments. It is specially designed to help researchers by offering labeled data for tasks like identifying humans in images, detecting actions over time, and understanding scenes. Tests using this dataset show that adding thermal and depth data significantly improves the accuracy of identifying people and their actions. In his thesis, Christian Stippel also develops a generative technique to create trimodal datasets by translating RGB data into thermal and depth images using unsupervised learning. This generative method is a potential solution for situations where limited data is available or in environments with challenging conditions.

Congratulations to Christian Stippel for his outstanding scientific achievements!

Curious about our other news? Subscribe to our news feed, calendar, or newsletter, or follow us on social media.