Skip to content

Arkenbrien/RosbagClassifier

Repository files navigation

Rosbag Classifier

RosbagClassifier.m allows for classification of LiDAR data as gravel, chipseal, or grass using an RDF to examine arcs from seperate LiDAR channels. At the moment, 3 arcs from channels 2, 3, and 4 are used. Two reference points are used for extracting spacial features: Range from LiDAR PoI, and a RANSAC projected plane. Manually defined areas coresponding to the rosbag to be classified allow for determining the accuracy of the classification process. Accuracy results are plotted & saved to disk.

To run RosbagClassifier.m, the following is needed: Rosbag with LiDAR, GPS, and IMU data Manual Classification .mat file (see next section) RDFs specific to each channel (2, 3, & 5 are provided)

===========================================================================

Consecutive point cloud scans may be made using the lidar2Geo.m script found in the https://github.com/OUIDEAS/Cyberbag_LatLon_Plotter Repo.

This allows for creating of manually defined areas for use in scoring the classification results using the Manual_Classifier.m script.

===========================================================================

RDFs may be trained with scripts from the RDF_Creation_Scripts folder.

About

Classify a Rosbag using a Random Decision Forest

Resources

Stars

Watchers

Forks

Packages

No packages published

Languages