Latest Projects
Master Thesis: Explainable Transformer-based 3D Object Detection in LiDAR Point Clouds.
Recent advancements in explainability for 3D object detection have shown promise, especially for small-scale point clouds. However, existing methods struggle with larger point clouds common in autonomous driving and often introduce significant computational overhead, limiting real-time use. This work focuses on improving the real-time explainability of transformer-based 3D object detection models on LiDAR point clouds while maintaining low-latency performance. It also introduces a framework to visualize interpretability results alongside object detection in real time.
​
Thesis Advisor: Till Beemelmans, M.Sc.
​
Lab: Institut für Kraftfahrzeuge (ika) RWTH Aachen University
​
TerraWatch: Multimodal AI for Deforestation Detection
TerraWatch combines computer vision with multimodal LLM models to detect deforestation from satellite images and predict their causes and possible environmental effects.
This was a prototype that was built in 48 hours during the TUM AI Hackathon 2024. Further contributions are welcome.
​