Light Detection and Ranging (LiDAR) has been a popular remote sensing technology deployed in various applications, including land surveying, 3D object model acquisition and reconstruction, ancient archeological site exploration and so on.
In collaboration with McElhanney Ltd., our research group started the research on “LiDAR urban scene feature extraction” funded by Mitacs, Canada. We use LiDAR in an urban setting to capture and analyze large-scale 3D data, which precisely extract specific urban asset features from the point cloud (in 3D space) and the feature recognition process is supported with optical images if available. Our goal is to automatically detect and extract point features and linear features, to match those described in McElhanney’s Survey Code list. McElhanney uses the MITACS project outcome to help them produce CAD files delivered for engineering design and planning.
In contrast to capturing humans, vehicles and other commonly seen real-world objects, which are widely available for training machine learning techniques in applications like self-driving cars, creating LiDAR 3D datasets with ground truth for small urban assets useful for land surveying is challenging. In order to benefit the research community, we will continue update this page so that point-cloud models with ground truth can be available for researchers to further advance research in this direction.
We will continue to update this page as our research progresses.
By downloading data, either in whole or in part from our dataset, the user agrees that the data is only for research purposes, should not be used or repurposed for usages with commercial interests.
The user of the data bears full responsibilities on any outcome as a result of using the downloaded data.
Click here to access our publicly available dataset.