Point clouds from aerial and mobile mapping platforms are crucial in the geospatial domain. Recent research on Multi-View-Stereo-Matching and LiDAR sensor technology has led to data that is remarkably accurate, dense, and reliable. Beyond providing core geometric information, point clouds are foundational for tasks such as land use classification and 3D modeling. Recent advancements in geometric deep learning have accelerated the automatic interpretation of unordered point sets, making high-level 3D representations of the environment, like 3D building models, increasingly mainstream.

The 4th International Workshop on Point Cloud Processing convenes experts from industry, academia, and national mapping agencies. The workshop's primary focus is on the processing, evaluation, and interpretation of point clouds for mapping purposes.

Our group also participated in this interesting format of technological and scientific exchange. Prof. Dorota Iwaszczuk presented joint research outcomes by Qipeng Mei, Kevin Qiu, Dimitri Bulatov entitled: Leveraging domain knowledge to refine deep-learning-based semantic segmentation of outdoor point clouds.

Two half days in Stuttgart filled with cutting-edge research, innovative implementations and vivid discussions were inspiring, fostering collaboration and driving the field forward.

Link to the PCP website.