Exploring Open-Source Object Detection and Visualization in an Expanding Big Data Era

>>> REGISTER HERE! <<<

Speaker: Robert Dzur, Sr. Vice President and Systems Manager, Spatial Data, Bohannan Huston, Inc.

Hosted by: Dr. Vaclav Petras, CGA Faculty Fellow and Senior Researcher

Summary: In this interactive studio, participants will apply an object detection model to identify remnant buildings in LiDAR-derived digital elevation models. Originally trained on a New Mexico airborne LiDAR dataset of high desert and urban landscapes, participants will apply the model to a subset of North Carolina Phase 3 2015 LiDAR data near Siler City. A set of Jupyter Notebooks are used to facilitate raster data preparation, point cloud reprojection, filtering and colorization for visualization. Methods and tools covered include object detection using YOLOv8 (You Only Look Once), point cloud visualization in Entwine, feature extraction using GRASS GIS, and displaying raster, vector and point cloud data outputs in QGIS. This hands-on studio can appeal to anyone interested in object recognition, point cloud processing, data accessibility and interpretation.

About the speaker: Robert Dzur is a Senior Vice President and Systems Manager of Spatial Data at Bohannan Huston, Inc. (BHI) where for the last 20 years he has led a variety of GIS, remote sensing, photogrammetric and LiDAR mapping projects. His current professional interests include high precision geospatial dataset development, historical mapping projects, and open-source technologies, and he is proficient in commercial and open-source GIS environments such as GRASS GIS, PDAL, and Entwine. Rob did his undergraduate work in Spanish at the University of Arkansas where he also completed his M.A. in Geography. Before joining BHI, Rob worked at the Center for Advanced Spatial Technologies (Univ. of Arkansas) and then spent 4 years in Bolivia as an international technical consultant for the World Bank, managing cadastral mapping projects.

Recording: https://youtu.be/tXUzw09kQus