Driton Salihu

Chair of Media Technology
Room 2934, Main Campus TUM
Arcisstr. 21, 80333 München
Hey! I am a Ph.D. candidate at the Chair of Media Technology at the Technical University of Munich, advised by Prof. Eckehard Steinbach, currently working on 6G-life project.
I completed my Master’s degree in Computer Engineering and Electrical Engineering at the Technical University of Munich. During my studies I participated in many different projects. For my Master’s thesis I was supervised by Dr. Christoph Bachhuber and Prof. Eckehard Steinbach, where I worked on modern illumination estimation for Augmented Reality. I was also fortunate to collaborate with Prof. Walter Stechele, Dr. Nael Fasfous, and Dr. Manoj Rohit Vemparala throughout the entire duration of my Master’s exploring the intricate nature of quantization and pruning in Deep Neural Networks.
I earned my Bachelor’s degree in Computer Engineering and Electrical Engineering from the Technical University of Munich, supervised by Prof. Eckehard Steinbach and Dr. Rania Hassen.
While pursuing my degrees I gained practical experience at companies such as BMW Group, Nokia, and Bertrandt (formerly Philotech GmbH).
My research interests lie at the intersection of computer vision and machine learning, with a particular focus on 3D vision and its applications as well as efficient AI (specifically quantization and pruning). I am currently working on 3D scene reconstruction (from images and/or point clouds), 3D pose estimation, 3D retrieval, and representation learning. Additionally, I am grateful that during my PhD, I had the opportunity to collaborate with Prof. Zhi Jin through multiple research visits to Shenzhen, China
Outside of research, I have a strong interest in tennis, reading, and baking. Feel free to reach out if you’d like to play a match!
I am currently seeking research opportunities outside of university, so don’t hesitate to contact me via email!
selected publications
- ZeroCAD: Leveraging 2D Outpainting as Guidance for Zero-Shot 3D Instance-based Scene ReconstructionIn Under Review, 2025
- DeepSPF: Spherical SO (3)-Equivariant Patches for Scan-to-CAD EstimationIn The Twelfth International Conference on Learning Representations (ICLR2024), 2024
- SGPCR: Spherical Gaussian Point Cloud Representation and its Application to Object Registration and RetrievalIn The 2023 IEEE/CVF Winter Conference on Applications of Computer Vision (WACV 2023), Jan 2023
- Hardware-aware mixed-precision neural networks using in-train quantizationIn British machine vision conference (BMVC), Jan 2021