Full Professor, Technical University Munich
3 papers at NeurIPS 2025
We present OrthoLoC, a large-scale UAV localization dataset pairing aerial imagery with orthographic geodata, enabling a new localization paradigm. Our AdHoP technique enhances matching-based localization pipelines.
learning pixel-dense multi-task representations from apparent motion
IPFormer introduces context-adaptive instance proposals for vision-based 3D Panoptic Scene Completion, outperforming the state-of-the-art and reducing runtime by over 14x.
