README UPDATED ON COMMIT ()

Usage:

• Open in Visual Studio 2017
• $config = project common/GConfig.cs

• Create a folder with 4 subfolders: lidar, dmr, tests and augmentation.
• in $config, change the workspace dirs variables in region [global]
• in $config, change the tool paths to point to your resources folder

Result: las files and dmr files in /dmr, /lidar directories of your workspace

• In $config, under [downloader] section, set the execution type
  • They are self explanatory, define the elements you want processed bellow in the downloader section.
• Rebuild project, set downloader as startup project and start the program.

Result: files able to be viewed/labeled in PointCloudia, create augmentables from. 4 files. saved in workspace/lidar/:

• .txt point cloud file

• .txt class file (class labels for each point)

• .txt intensity file (intensity labels for each point)

• .pcd RBNN result file

• In $config, set the RBNN_R_VALUES parameter to include all the RBNN values you want to obtain.

• Set the executor project as startup project and run it.

Result: files filled with augmentables in workspace/augmentation

• In $config set the following:

  • candidateContextRadius - a candidate has a context of some radius. This is important because it will determine the upper Z bound of sampling augmentables. (We don't want to sample 500 meters above the highest point of the terrain because the context will always be empty and the example, therefore, trivial)
  • ObjectsToAdd - how many objects do you want to create
  • AUGMENTATION_RBNN_R_SPAN - for determining the distance between the augmentable and the closest point in the dataset. Should be between a small r and the candidateContextRadius. The denser you make it, the more exact it will be, but the longer it will take to compute.
  • GetAugmentableObjectPalette() - definitions for possible floating objects that will be randomly sampled.

• Set augmentation_sampler project as startup project and run the program.

for each augmentable:

• [int id] [X,Y,Z vec float position] [X,Y,Z vec float scale] [string shape name] [X,Y,Z vec float airplane position] [float distance from transitive ground]\n

Viewing data in PointCloudia

Laz preprocessor is used to transform .laz files into other file forms and also extract additional data if needed. It can:

• Add RGB values for points in ARSO datasets (Using Slemenik's application)
• Output xyzRGB text files
• Output class text files - each line is a number which determines a point's class
• Output floating object class files - using the RBNN algorithm

• You need to have the tool las2txt in your path. You can check the LASTools project to get it.

• To acquire the initial data you have two options:
  • Use Slemenik's tool to download RGB enriched .laz files
  • Have a set of .laz files in a directory prepared
• Configure the core project via the CoreProcedure.cs file's variables in the config region
  • set the path to your rbnn.exe file (it is included in the core project in the resources folder)
  • set the desired radius values you would like rbnn to execute for
• Configure the executor project via the Program.cs file's variables
  • set the directory in which the .laz files you want to preprocess are contained
• Run the executor project and wait for the program to finish
• You will now have all your desired files residing by the .laz files in the same folder