Thank you for downloading the Semantic Urban Mesh (SUM) benchmark datasets.

Citation: TBD

Files in the folder: 
1. SUM_Helsinki_C6_mesh
	1.1 Files: 
		- mesh files (*.ply)
		- texture images (*.jpg)	
	1.2 Extended ply file properties in our dataset:
		- semantic lables are listed in the header comments, they are defined as: 
			label -1 unclassified    ,RGB: 0   0   0 
			label 1 ground (terrain) , RGB: 170 85  0          
			label 2 (high) vegetation, RGB: 0   255 0 
			label 3 building         , RGB: 255 255 0 
			label 4 water            , RGB: 0   255 255 
			label 5 car              , RGB: 255 0   255 
			label 6 boat             , RGB: 0   0   153
		- face_segment_id and label_probabilities were used for annotation, the user 
		  can visualize them in our annotation software (https://github.com/tudelft3d/3D_Urban_Mesh_Annotator). 
		  
2. SUM_Helsinki_C6_pcl_30
	2.1 Files:
		- sampled point clouds (*.ply)
	2.2 Density:
	    - around 30 pts / m2 (refer to surface area)
	2.3 Extended ply file properties in our dataset:
		- points_face_belong_id indicates the corresponding face of each point.

3. SUM_Helsinki_C6_pcl_300
	2.1 Files:
		- sampled point clouds (*.ply)
	2.2 Density:
	    - around 300 pts / m2 (refer to surface area)
	2.3 Extended ply file properties in our dataset:
		- points_face_belong_id indicates the corresponding face of each point.