semantic-8

semantic-8 is a benchmark for classification with 8 class labels, namely {1: man-made terrain, 2: natural terrain, 3: high vegetation, 4: low vegetation, 5: buildings, 6: hard scape, 7: scanning artefacts, 8: cars}. An additional label {0: unlabeled points} marks points without ground truth and should not be used for training! In total over a billion points are provided. Please check out the reduced benchmark if your method is too computational demanding for the full data set.

Jump to download

Training Set

PreviewName Number of Points SceneDescription
1bildstein129302501ruralchurch in bildstein
2bildstein323765246ruralchurch in bildstein
3bildstein524671679ruralchurch in bildstein
4domfountain135494386urbancathedral in feldkirch
5domfountain235188343urbancathedral in feldkirch
6domfountain335049972urbancathedral in feldkirch
7untermaederbrunnen116658648ruralfountain in balgach
8untermaederbrunnen319767991ruralfountain in balgach
9neugasse50109087urbanneugasse in st. gallen
10sg27_1161044280ruralrailroad tracks
11sg27_2248351425urbantown square
12sg27_4280994028ruralvillage
13sg27_5218269204suburbancrossing
14sg27_9222908898urbansoccer field
15sg28_4258719795urbantown square

Test Set

PreviewName Number of Points SceneDescription
1stgallencathedral128181979urbancathedral in st. gallen
2stgallencathedral331328976urbancathedral in st. gallen
3stgallencathedral632342450urbancathedral in st. gallen
4marketsquarefeldkirch123228738urbanmarket square in feldkirch
5marketsquarefeldkirch422760334urbanmarket square in feldkirch
6marketsquarefeldkirch723264911urbanmarket square in feldkirch
7birdfountain136627054urbanfountain in feldkirch
8castleblatten1152248025ruralcastle in blatten
9castleblatten5195356302ruralcastle in blatten
10sg27_3422445052suburbanhouses
11sg27_6226790878urbancity block
12sg27_8429615314urbancity center
13sg27_10285579196urbantown square
14sg28_2170158281ruralfarm
15sg28_5269007810suburbanbuildings


Download

We provide both training and test data as zipped ascii text files with format {x, y, z, intensity, r, g, b}. The ground truth is provided as single column ascii file, where the row ids of the class labeles and the points correspond. 7zip was used for compression and is available for both windows and linux.

point clouds for testing as zipped ascii files:
birdfountain_station1_xyz_intensity_rgb.7z( 0.25 GB )
castleblatten_station1_intensity_rgb.7z( 0.24 GB )
castleblatten_station5_xyz_intensity_rgb.7z( 0.70 GB )
marketplacefeldkirch_station1_intensity_rgb.7z( 0.17 GB )
marketplacefeldkirch_station4_intensity_rgb.7z( 0.15 GB )
marketplacefeldkirch_station7_intensity_rgb.7z( 0.15 GB )
sg27_station10_intensity_rgb.7z( 1.56 GB )
sg27_station3_intensity_rgb.7z( 2.40 GB )
sg27_station6_intensity_rgb.7z( 1.27 GB )
sg27_station8_intensity_rgb.7z( 2.08 GB )
sg28_station2_intensity_rgb.7z( 0.94 GB )
sg28_station5_xyz_intensity_rgb.7z( 1.35 GB )
stgallencathedral_station1_intensity_rgb.7z( 0.22 GB )
stgallencathedral_station3_intensity_rgb.7z( 0.22 GB )
stgallencathedral_station6_intensity_rgb.7z( 0.22 GB )
point clouds for training as zipped ascii files:
bildstein_station1_xyz_intensity_rgb.7z( 0.20 GB )
bildstein_station3_xyz_intensity_rgb.7z( 0.17 GB )
bildstein_station5_xyz_intensity_rgb.7z( 0.18 GB )
domfountain_station1_xyz_intensity_rgb.7z( 0.28 GB )
domfountain_station2_xyz_intensity_rgb.7z( 0.25 GB )
domfountain_station3_xyz_intensity_rgb.7z( 0.23 GB )
neugasse_station1_xyz_intensity_rgb.7z( 0.32 GB )
sg27_station1_intensity_rgb.7z( 1.87 GB )
sg27_station2_intensity_rgb.7z( 2.72 GB )
sg27_station4_intensity_rgb.7z( 1.59 GB )
sg27_station5_intensity_rgb.7z( 1.25 GB )
sg27_station9_intensity_rgb.7z( 1.22 GB )
sg28_station4_intensity_rgb.7z( 1.40 GB )
untermaederbrunnen_station1_xyz_intensity_rgb.7z( 0.17 GB )
untermaederbrunnen_station3_xyz_intensity_rgb.7z( 0.17 GB )
ground truth labels for training as zipped ascii files ( 0.01 GB)